DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-20 are pending in the present application.
Claim Objections
Claims 7, 9, 17-18 are objected to because of the following informalities: he limitation “…to the update the local ML…” contains a minor grammatical error. The examiner recommends rephrasing the limitation as “..to
Appropriate correction is required.
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-20 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1-20 of U.S. Patent No. 11,966,818. Although the claims at issue are not identical, they are not patentably distinct from each other because the instant application is a broader version of Patent 11,966,818’s inventive concept, which is directed to determining a way for a machine-learning node in a blockchain network to recover after a fault and safely rejoin training. Therefore, it is not patentably distinct from the Patent.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim 20 is rejected under 35 USC §101 as being directed to non-statutory subject matter. The claim does not fall within at least one of the four categories of patent eligible subject matter because the term “non-transitory media” has been redefined by the applicant to encompass signals per se ([0081] “The term "non-transitory media," and similar terms, as used herein refers to any media that store data and/or instructions that cause a machine to operate in a specific fashion…” [emphasis added]). See MPEP 2106.03 (I), MPEP 2106 (II) “In Mentor Graphics, the court interpreted the claims in light of the specification, which expressly defined the medium as encompassing "any data storage device" including random-access memory and carrier waves. Although random-access memory and magnetic tape are statutory media, carrier waves are not because they are signals similar to the transitory, propagating signals held to be non-statutory in Nuijten. 851 F.3d at 1294, 112 USPQ2d at 1133 (citing In re Nuijten, 500 F.3d 1346, 84 USPQ2d 1495 (Fed. Cir. 2007)). Accordingly, because the BRI of the claims covered both subject matter that falls within a statutory category (the random-access memory), as well as subject matter that does not (the carrier waves), the claims as a whole were not to a statutory category and thus failed the first criterion for eligibility.” As such, the claim “A non-transitory machine-readable storage medium comprising…” is seen as encompassing signals per se. The examiner recommends amending the specification to avoid directing the claim towards non-transitory media.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over US 2019/0089716 A1 to Stöcker (hereinafter Stöcker, as submitted in IDS dated 12/14/2023) in view of “Distributed Key Generation in the Wild” to Kate et al (hereinafter Kate, as submitted in IDS dated 12/14/2023), further in view of US 2019/0287026 A1 to Calmon et al (hereinafter Calmon, as submitted in IDS dated 12/14/2023).
As per claim 1, Stöcker teaches a self-healing computer node of a blockchain network comprising a plurality of computing nodes, the self-healing computer node recovering from a fault condition within the blockchain network and being programmed to ([0073] “In a particularly preferred embodiment of the present peer-to-peer network, the peer-to-peer application can be a block chain or decentral ledger comprising at least two blocks coupled to each other (e.g. Ethereum Block chain with Smart Contracts). The block chain technology or “decentral ledger technology” is already used in the payment by means of a crypto currency, such as Bitcoin...” [0143] “In a first step 801, e.g. node 404.2 may detect, by using the detecting means 458.2, that an operation error of the node 404.2 has occurred. By way of example, the detecting means 458.2 detect a reboot process of the node 404.2. Such a reboot process may be an (indirect) indication of a previously occurred operation error, such as a power outage or a communication connection interruption with the other nodes 402, 404.1 of the peer-to-peer network 400.” Examiner Note: Stöcker teaches a node (404.2) in a blockchain network with inbuilt error detection.)
generate a first blockchain transaction comprising an indication that the self-healing computer node is out-of-sync with a most recent iteration, wherein the first blockchain transaction is to be added to a distributed ledger and informs the plurality of computing nodes that the self-healing computer node is not ready to participate ([0073] “In a particularly preferred embodiment of the present peer-to-peer network, the peer-to-peer application can be a block chain or decentral ledger comprising at least two blocks coupled to each other (e.g. Ethereum Block chain with Smart Contracts). The block chain technology or “decentral ledger technology” is already used in the payment by means of a crypto currency, such as Bitcoin. It has been recognized that by a particular configuration of a block chain, data stored or to be stored on the block chain can be checked without the need of a central server. In addition, the block chain can be used to generate messages, transaction agreements between at least two entities and the like. The generation can be caused by at least one peer-to-peer module in a tamper-proof manner. The block chain according to the present embodiment is particularly a decentralized, peer-to-peer-based register in which all data related to an action and other messages sent by peer-to-peer modules can be logged.” [0143] “In a first step 801, e.g. node 404.2 may detect, by using the detecting means 458.2, that an operation error of the node 404.2 has occurred. By way of example, the detecting means 458.2 detect a reboot process of the node 404.2. Such a reboot process may be an (indirect) indication of a previously occurred operation error, such as a power outage or a communication connection interruption with the other nodes 402, 404.1 of the peer-to-peer network 400.” [0144] “In a next step 802, the evaluating means 460.2 check whether a data synchronization of the data stored in the peer-to-peer application 424 and on the node 404.2 is required. Preferably, each data (set) stored in the peer-to-peer application 424 may be associated with a time stamp. The evaluating means 460.2 may compare the time stamp of the last stored data with the current time. If, e.g. the determined time period between said time stamp and the current time exceeds a preset threshold period, the evaluating means 460.2 may come to the result that data synchronization is required.” [0146] “As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400.” Examiner Note: Stöcker’s peer-to-peer network is equivalent to a blockchain network, per 0073. As such, Stöcker’s node 404.2 sending a message to other nodes of the network indicating an operation error is seen as equivalent to sending an indication that the node is not ready to participate in the network. Further, 0073 indicates that the message is added to a distributed ledger.);
obtain a global state from the distributed ledger ([0037] “Alternatively or additionally, the evaluating means can compare the last stored data on said node with the last stored data on the further nodes. A discrepancy between the stored data may be an indication that data synchronization is required. If a discrepancy is not determined data synchronization may not be required.” [0073] “In a particularly preferred embodiment of the present peer-to-peer network, the peer-to-peer application can be a block chain or decentral ledger comprising at least two blocks coupled to each other…” Examiner Note: The information from other nodes in the network is seen as equivalent to the state of the distributed ledger.);
compare the obtained global state with a local state at the self-healing computer node to determine whether the local state is consistent with global state ([0144] “In a next step 802, the evaluating means 460.2 check whether a data synchronization of the data stored in the peer-to-peer application 424 and on the node 404.2 is required. Preferably, each data (set) stored in the peer-to-peer application 424 may be associated with a time stamp. The evaluating means 460.2 may compare the time stamp of the last stored data with the current time. If, e.g. the determined time period between said time stamp and the current time exceeds a preset threshold period, the evaluating means 460.2 may come to the result that data synchronization is required.”);
upon determining that the local state is not consistent with the global state, trigger a corrective action using the blockchain network to recover the local state to be consistent with the global state ([0145] “Upon the detection of the necessity of data synchronization, in a next step 803 the initiating means 462.2 may initiate the data synchronization process by transmitting a data synchronization indication to the communication module 430.2 of the node 404.2.” [0146] “Preferably, directly upon receipt of the data synchronization indication the communication module 430.2 may start/trigger one or more data synchronization action(s) (step 804). As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400…”).
However, Stöcker does not explicitly teach:
generate a second blockchain transaction comprising an indication that the self-healing computer node is in-sync with the most recent iteration of training a machine-learned model, wherein the second blockchain transaction is to be added to the distributed ledger and informs the plurality of computing nodes that the self-healing computer node is ready to participate in a subsequent iteration of training the machine-learned model; and
re-enrolling with the blockchain network to participate in the subsequent iteration of training the machine-learned model.
Kate teaches, in an analogous system, generate a second blockchain transaction comprising an indication that the self-healing computer node is in-sync with the most recent iteration, wherein the second blockchain transaction is to be added to the distributed ledger and informs the plurality of computing nodes that the self-healing computer node is ready to participate in a subsequent iteration (Page 12, “As n − t − f ≥ (n+t+1)/2 for n ≥ 3t + 2f + 1, every honest finally up node Pj will send ready message (Pd, τ, ready, C, aj (m)) to every system node Pm as either the received echo messages are greater than required bound (d n+t+1 2 e) or it has already received t + 1 ready messages.” Page 10, “verify-poly(C, i, a) verifies that the given polynomial a of Pi is consistent with the commitment C. Here, a(y) = Pt`=0 a`y` is a degree-t polynomial. The predicate is true if and only if g a` = Qt j=0(Cj`) ij for all ` ∈ [0, t].” Examiner Note: The variables in Kate’s ready message above correspond to Pd= network ‘dealer’ identification, τ = session identification, C = commitment, and aj(m) = verification value. Thus, the message is an indication to the network that it is ready, and offers proof of both synchronization and trustworthiness in the form of session identification, commitment, and verification value, as if any of these values are invalid, the system will reject participation of the node. The combination of Kate’s blockchain security method including a ready message into Stöcker’s blockchain node synchronization method would obviously result in Stöcker’s node sending a ready message as part of the node’s return to normal operation. That is, Stöcker’s self-healing node would generate a second blockchain transaction comprising an indication that the self-healing computer node is in-sync with the most recent iteration, wherein the second blockchain transaction is to be added to the distributed ledger and informs the plurality of computing nodes that the self-healing computer node is ready to participate in a subsequent iteration.); and
re-enrolling with the blockchain network to participate in the subsequent iteration (Page 12, “As all ready messages will be eventually received by the finally up nodes according to Lemma 4.1, each finally up honest node will receive at least n − t − f verifiably correct ready messages. Consequently, each honest finally up node will complete the protocol Sh by outputting (Pd, τ,shared) messages.” Examiner Note: Once Kate’s node receives indications from the other nodes participating in their network (e.g., that they are also ready and trustworthy), the node executes an operation in correspondence with the network. When applied to Stöcker’s node and network, the result would be that Stöcker’s newly synchronized node participates (i.e., is enrolled in) the operation of Stöcker’s network following the node’s sending of a ready message to the network).
Stöcker and Kate are analogous art because they are both directed towards cryptographically secured networks. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker’s blockchain node with Kate’s node verification. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the security of the network, which can be accomplished by adapting the system to include node verification (Kate, Introduction, “Numerous online cryptographic applications require a trusted authority to hold a secret. However, this requirement always leads to the problem of single point of failure and sometimes to the more undesirable problem of key escrow. Solving these problems is of paramount importance while designing systems over the Internet where denial-of-service attacks and malicious entities are widespread. A distributed key generation (DKG) [42] protocol overcomes these problems using a complete distribution of the trust among a set of servers.”).
Stöcker teaches a self-healing blockchain node and network, generating a blockchain transaction when a node is out-of-sync with the network, a distributed ledger resynchronizing the node to the global state of the network, and returning to normal operation, but does not specifically disclose a decentralized machine learning system.
Calmon teaches A system of decentralized machine learning (ML) comprising ([0021] Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data).
Stöcker teaches generating a blockchain transaction indicating that a node in the network is out-of-sync with the network and unready to participate in the network, and adding that transaction to the distributed ledger, but does not explicitly disclose the network being used for training a machine-learned model.
Calmon teaches training a machine-learned model, participate in a subsequent iteration of training the machine-learned model ([0032] “The learning service providing node 120 generates a learning model based on the training data and provides the learning model to the verification nodes 130.” [0048] “In response to receiving respective training data, both the LSP node A 204 and the LSP node B 206 may build or otherwise compose learning models based on the training data.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, Stöcker’s node would indicate that it is specifically not ready to participate in ML model training, and the resulting combination would generate a first blockchain transaction comprising an indication that the self-healing computer node is out-of-sync with a most recent iteration of training a machine-learned model, wherein the first blockchain transaction is to be added to a distributed ledger and informs the plurality of computing nodes that the self-healing computer node is not ready to participate in a subsequent iteration of training the machine-learned model.);
Stöcker teaches obtaining a global state from the distributed ledger, but does not disclose an ML state.
Calmon teaches an ML state ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system, the resulting system specifically obtains a global ML state from the distributed ledger).
Stöcker teaches comparing the obtained global state with a local state at the self-healing computer node to determine whether the local state is consistent with global state, but does not disclose comparing ML states.
Calmon teaches an ML state ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system, the resulting system specifically obtains a global ML state from the global ledger (e.g., the model being passed node to node between iterations), and compares it to the local ML state.);
Stöcker teaches upon determining that the local state is not consistent with the global state, trigger a corrective action using the blockchain network to recover the local state to be consistent with the global state, but does not disclose an ML state.
Calmon teaches an ML state ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system, the resulting system specifically obtains a global ML state from the global ledger (e.g., the model being passed node to node between iterations), and compares it to the local ML state.);
The combination of Stöcker and Kate teaches generate a second blockchain transaction comprising an indication that the self-healing computer node is in-sync with the most recent iteration, wherein the second blockchain transaction is to be added to the distributed ledger and informs the plurality of computing nodes that the self-healing computer node is ready to participate in a subsequent iteration, but does not disclose training a machine learning model.
Calmon teaches training a machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, Stöcker’s node would indicate that it is specifically ready to participate in ML model training, and the resulting combination would generate a second blockchain transaction comprising an indication that the self-healing computer node is in-sync with the most recent iteration of training a machine-learned model, wherein the second blockchain transaction is to be added to the distributed ledger and informs the plurality of computing nodes that the self-healing computer node is ready to participate in a subsequent iteration of training the machine-learned model); and
The combination of Stöcker and Kate discloses re-enrolling with the blockchain network to participate in the subsequent iteration, but does not disclose training the machine-learned model.
Calmon teaches training the machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically re-enrolls itself in training the ML model.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 2, the combination of Stöcker, Kate, and Calmon as detailed above teaches The system of claim 1.
Stöcker teaches further comprising: a master node selected from among the participant nodes in most recent iteration ([0137] FIG. 4 shows a further embodiment of a peer-to-peer network 400 according to the present invention. The depicted peer-to-peer network 400 comprises three nodes 402, 404.1, 404.2. For instance, two second (slave) nodes 404.1, 404.2 and one first (master) node 402 can be provided. However, in the present embodiment, it is not necessary that the nodes 402, 04.1, 404.2 are capable of conducting the above described time synchronization process. [0143] “In a first step 801, e.g. node 404.2 may detect, by using the detecting means 458.2, that an operation error of the node 404.2 has occurred. By way of example, the detecting means 458.2 detect a reboot process of the node 404.2. Such a reboot process may be an (indirect) indication of a previously occurred operation error, such as a power outage or a communication connection interruption with the other nodes 402, 404.1 of the peer-to-peer network 400.” [0144] “In a next step 802, the evaluating means 460.2 check whether a data synchronization of the data stored in the peer-to-peer application 424 and on the node 404.2 is required. Preferably, each data (set) stored in the peer-to-peer application 424 may be associated with a time stamp. The evaluating means 460.2 may compare the time stamp of the last stored data with the current time. If, e.g. the determined time period between said time stamp and the current time exceeds a preset threshold period, the evaluating means 460.2 may come to the result that data synchronization is required.” [0146] “As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400.”).
Stöcker teaches a master node selected from among the participant nodes in most recent iteration, but does not explicitly teach participant nodes from the plurality of nodes on the blockchain network, wherein the participant nodes were enrolled to participate in the most recent iteration of training a machine-learned model, and training the machine-learned model.
Calmon teaches participant nodes from the plurality of nodes on the blockchain network, wherein the participant nodes were enrolled to participate in the most recent iteration of training a machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: Calmon’s nodes accepting the last iteration of an ML model and then improving on said model is seen as equivalent to participating in the most recent iteration of training a machine-learned model.); and
training the machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically selects a master node from those participating in the most recent iteration of training the machine-learned model.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 3, the combination of Stöcker, Kate, and Calmon as detailed above teaches The system of claim 2.
Stöcker teaches wherein the master node is programmed to: receive the indication that the self-healing computer node is out-of-sync with a most recent iteration ([0143] “In a first step 801, e.g. node 404.2 may detect, by using the detecting means 458.2, that an operation error of the node 404.2 has occurred. By way of example, the detecting means 458.2 detect a reboot process of the node 404.2. Such a reboot process may be an (indirect) indication of a previously occurred operation error, such as a power outage or a communication connection interruption with the other nodes 402, 404.1 of the peer-to-peer network 400.” [0144] “In a next step 802, the evaluating means 460.2 check whether a data synchronization of the data stored in the peer-to-peer application 424 and on the node 404.2 is required... The evaluating means 460.2 may compare the time stamp of the last stored data with the current time. If, e.g. the determined time period between said time stamp and the current time exceeds a preset threshold period, the evaluating means 460.2 may come to the result that data synchronization is required…” [0145] “Upon the detection of the necessity of data synchronization, in a next step 803 the initiating means 462.2 may initiate the data synchronization process by transmitting a data synchronization indication to the communication module 430.2 of the node 404.2.” [0146] “Preferably, directly upon receipt of the data synchronization indication the communication module 430.2 may start/trigger one or more data synchronization action(s) (step 804). As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400…” [0149] “In step 805, the further node 404.1 may receive the data synchronization request message.” Examiner Note: Stöcker’s synchronization request is sent to master node 402.).
Stöcker teaches receive the indication that the self-healing computer node is out-of-sync with a most recent iteration but does not explicitly teach training the machine-learned model; and excludes the self-healing computer node from participating in a subsequent iteration of training the machine-learned model based in the indication that the self-healing computer node is out-of-sync such that training parameters associated with the local ML state of the self- healing computer node are prevented from being applied to the machine-learned model.
Calmon teaches training the machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically receives an indication that the node is out-of-sync with a most recent iteration of training the machine-learned model (e.g. the timestamp of the local node’s model differs from the timestamp of the global model.).); and
excludes the self-healing computer node from participating in a subsequent iteration of training the machine-learned model based in the indication that the self-healing computer node is out-of-sync such that training parameters associated with the local ML state of the self-healing computer node are prevented from being applied to the machine-learned model ([0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes. The verification nodes verify if the improvement of the model in this turn satisfies the smart contract, and if so, the next participant gets a turn to improve the model. This process ends (in one embodiment) once the model cannot be improved by one or more participants; in an alternative embodiment once a model cannot be improved by a participant, that participant no longer participates in the iterative process and no longer receives updated learning models” [0058] “In some embodiments, the method 500A of FIG. 5A may further include one or more of iteratively receiving a learning model from one or more of a plurality of learning service providers, executing the iteratively received learning model on one or more of a plurality of test data sets provided by one or more of a plurality of data providers to verify whether the iteratively received learning model satisfies a predetermined performance threshold, selecting one or more learning service providers to further update the learning model, based on the execution and verification from the current and previous iterations, the updated learning model to be communicated to the processor at the start of the next iteration, and terminating the process based on the execution and verification steps from the current and previous iteration.” [0048] “In 236, the verification node 208 determines whether each learning model satisfies performance requirements as implemented via a smart contract.” Examiner Note: When Stöcker’s node reset is applied to Calmon’s Decentralized ML system selecting a node for participation, the resulting system would refrain from selecting the resetting node to update the ML model, as it would have communicated to the system that it would be unable to participate in the next iteration of the model training (see claim 1: “…generate a first blockchain transaction…”.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 4, the combination of Stöcker, Kate, and Calmon as detailed above teaches The system of claim 3.
Stöcker teaches wherein excluding the self-healing computer node from participating in a subsequent iteration enables tolerate the fault condition within the blockchain network ([0013] “A further issue of prior art peer-to-peer network is that after an operation error of a node, such as an interruption of the power supply of the node, the data stored in the peer-to-peer application and on the node with the error may differ from the data stored in the peer-to-peer application and on the further nodes of the peer-to-peer network. Thereby, it is a general concern to restore the full functionality of the node as soon as possible. In particular, a fast data synchronisation [sic] of the data stored in the peer-to-peer application and on the node with the error and the data stored in the peer-to-peer application and on the further nodes of the peer-to-peer network is required.” [0034] “According to a particularly preferred embodiment of a peer-to-peer network of the present invention, the at least one node may comprise at least one controlling module having at least one detecting means configured to detect at least one operation error of the node. The controlling module may comprise at least one evaluating means configured to evaluate whether the detected operation error requires data synchronization of the data stored in the peer-to-peer application of (and on) the node. The controlling module may comprise at least one initiating means configured to initiate a data synchronization action if data synchronization is required.” [0068] “The peer-to-peer application can be built upon the following elements: PEER-TO-PEER network, Consensus System/Protocol, Data Structure, Merkle Trees, Public Key Signatures, Byzantine Fault Tolerance. It replicates data based on a consensus principle. It is auditable and traceable. Each of these systems may be comprised by a peer-to-peer network.” Examiner Note: The examiner sees the error tolerance and further byzantine fault tolerance of Stöcker’s node and network as equivalent to tolerating a fault condition within the network.).
Stöcker teaches tolerating fault conditions within the blockchain network, but does not explicitly teach training the machine-learned model.
Calmon teaches training the machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically prevents out-of-sync nodes from participating in ML training.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 5, the combination of Stöcker and Calmon as detailed above teaches The system of claim 2.
Stöcker teaches wherein the master node is programmed to: receive the indication that the self-healing computer node is in-sync with a most recent iteration ([0137] “FIG. 4 shows a further embodiment of a peer-to-peer network 400 according to the present invention. The depicted peer-to-peer network 400 comprises three nodes 402, 404.1, 404.2. For instance, two second (slave) nodes 404.1, 404.2 and one first (master) node 402 can be provided.” [0146] “As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400.” [0155] “After the entire data has been synchronized the bandwidth of the connection 428 can be reset and the nodes 402, 404.1, 404.2 and/or the peer-to-peer network can be further operated according to a regular modus.” Examiner Note: When Kate’s ready message as detailed in claim 1 (“…generate a second…” “re-enrolling with the blockchain…”) is combined with Stöcker’s node synchronization including a master node, Stöcker’s master node would receive a ready indication from the self-healing node once synchronization is complete.); and
includes the self-healing computer node for participating in the subsequent iteration based in the indication that the self-healing computer node is in-sync such that parameters associated with the local state of the self- healing computer node are applied ([0155] “After the entire data has been synchronized the bandwidth of the connection 428 can be reset and the nodes 402, 404.1, 404.2 and/or the peer-to-peer network can be further operated according to a regular modus.” Examiner Note: The examiner sees the network (and specifically resetting node 404.2) operating according to a regular modus as necessarily including the re-synchronized node 404.2 in subsequent iterations (as otherwise, node 404.2 would not be operating normally).).
Stöcker teaches wherein the master node is programmed to: receive the indication that the self-healing computer node is in-sync with a most recent iteration; and includes the self-healing computer node for participating in the subsequent iteration based in the indication that the self-healing computer node is in-sync such that parameters associated with the local state of the self- healing computer node are applied, but does not explicitly teach training the machine-learned model; and training parameters associated with the local ML state applied to the machine-learned model.
Calmon teaches training the machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically receives an indication that the node is in sync with a most recent iteration of training the machine-learned model.); and
training parameters associated with the local ML state applied to the machine-learned model ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” [0030] “As described herein, a learning model is a parameterized piece of code or mathematical function. A trained learning model is the learning model plus a set of parameter values. For example, a specific neural net architecture implementation plus values of the neural net weights. In these examples, a trained learning model takes data as an input, and produces an output such as a cluster-set, a prediction, a class-label, a regression value, a recommended action, or the like, as output (other types of responses/outcomes are possible as well).” [0033] “FIG. 1B illustrates a mutual learning environment 100B where multiple nodes provide both data and learning models in a mutual exchange of value to the other. In this example, the same process is performed as described in FIG. 1A except that there are two learning service providing nodes 121 and 122 with their own data. In this example, both learning service providing nodes 121 and 122 provide data to the verification nodes 130 which partition the data, hash the partition data (test/train) from each, and store the hashed data via the blockchain. In addition, the verification nodes 130 provide the training data from the learning service providing node 121 to the learning service providing node 122, and provide the training data from the learning service providing node 122 to the learning service providing node 121.” Examiner Note: Per 0030, Calmon’s ML model is seen as including training parameters. Per 0033, Calmon’s node generates an ML model and passes the ML model from node to node. When Stöcker’s node and network are applied to Calmon’s decentralized ML system, the resulting system would specifically update the out-of-sync node with the most recent ML information including training parameters and a ML model.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 6, the combination of Stöcker, Kate, and Calmon as detailed above teaches The system of claim 5.
Stöcker does not explicitly teach wherein including the self-healing computer node for participating in the subsequent iteration of training the machine-learned model reintegrates the self-healing computer node in to the decentralized machine learning.
Calmon teaches The system of claim 5, wherein including the self-healing computer node for participating in the subsequent iteration of training the machine-learned model reintegrates the self-healing computer node in to the decentralized machine learning ([0058] “In some embodiments, the method 500A of FIG. 5A may further include one or more of iteratively receiving a learning model from one or more of a plurality of learning service providers, executing the iteratively received learning model on one or more of a plurality of test data sets provided by one or more of a plurality of data providers to verify whether the iteratively received learning model satisfies a predetermined performance threshold, selecting one or more learning service providers to further update the learning model, based on the execution and verification from the current and previous iterations, the updated learning model to be communicated to the processor at the start of the next iteration, and terminating the process based on the execution and verification steps from the current and previous iteration.” Examiner Note: When Stöcker’s node and network are applied to Calmon’s decentralized machine learning system including iteratively training a machine learning model, the resulting system would specifically re-enroll the node in the subsequent iteration of training the machine-learned model via decentralized machine learning.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 7, the combination of Stöcker, Kate, and Calmon as detailed above teaches The system of claim 2.
Stöcker teaches wherein the self-healing computer node is further programmed to: based on the triggering of the corrective action, obtain shared parameters generated by a participant node on the blockchain network ([0042] “In order to provide a particular fast data synchronization, according to a further embodiment the communication module of said node may be further configured to perform, upon receipt of the data synchronization indication, at least one data synchronization action from the group of data synchronization actions comprising: [0043] providing the data synchronization request message with a priority information…” [0046] “Further, data packets comprising the priority information may have priority in comparison with other data packets. Preferably, every data packet transmitted during a data synchronization process, such as data packets comprising at least a part of the requested data, can be provided with a priority information.” Examiner Note: Stöcker’s data synchronization of a node including receiving data packets from other nodes in the blockchain network is seen as equivalent to obtaining shared parameters generated by a participant node on the blockchain network, and Stöcker’s data synchronization indication is seen as equivalent to triggering corrective action.); and
apply the parameters to the update the local state at the self-healing computer node ([0151] “Upon receipt of the data packet(s) from the further node 404.1 the requesting node 404.2 may update its data of the peer-to-peer application (step 807).”).
Stöcker teaches wherein the self-healing computer node is further programmed to: based on the triggering of the corrective action, obtain shared parameters generated by a participant node on the blockchain network, wherein the shared parameters are based on a local participant node during the most recent iteration; and apply the parameters to the update the local state at the self-healing computer node but does not explicitly teach training parameters of a machine-learned model, and an ML state.
Calmon teaches training parameters, wherein the shared training parameters are based on a local model of the participant node being trained during the most recent iteration ([0030] “As described herein, a learning model is a parameterized piece of code or mathematical function. A trained learning model is the learning model plus a set of parameter values. For example, a specific neural net architecture implementation plus values of the neural net weights. In these examples, a trained learning model takes data as an input, and produces an output such as a cluster-set, a prediction, a class-label, a regression value, a recommended action, or the like, as output (other types of responses/outcomes are possible as well).” [0033] “FIG. 1B illustrates a mutual learning environment 100B where multiple nodes provide both data and learning models in a mutual exchange of value to the other. In this example, the same process is performed as described in FIG. 1A except that there are two learning service providing nodes 121 and 122 with their own data. In this example, both learning service providing nodes 121 and 122 provide data to the verification nodes 130 which partition the data, hash the partition data (test/train) from each, and store the hashed data via the blockchain. In addition, the verification nodes 130 provide the training data from the learning service providing node 121 to the learning service providing node 122, and provide the training data from the learning service providing node 122 to the learning service providing node 121.” Examiner Note: Per 0030, Calmon’s ML model is seen as including training parameters. Per 0033, Calmon’s node generates an ML model and passes the ML model from node to node. When Stöcker’s node and network are applied to Calmon’s decentralized ML system, the resulting system would specifically update the out-of-sync node with the most recent ML information including training parameters and a ML model.); and
the local ML state ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically updates the local ML state.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 8, the combination of Stöcker, Kate, and Calmon as detailed above teaches The system of claim 2.
Stöcker does not explicitly teach wherein the master node is further programmed to: obtain shared training parameters from the participant nodes on the blockchain network, wherein the shared training parameters are based on a local model of the participant nodes being trained during the most recent iteration; generate merged training parameters based on the shared training parameters; generate a transaction that includes an indication that the master node has generated the merged training parameters; cause the transaction to be written as a block on the distributed ledger.
Calmon teaches The system of claim 2, wherein the master node is further programmed to: obtain shared training parameters from the participant nodes on the blockchain network, wherein the shared training parameters are based on a local model of the participant nodes being trained during the most recent iteration ([0030] “As described herein, a learning model is a parameterized piece of code or mathematical function. A trained learning model is the learning model plus a set of parameter values. For example, a specific neural net architecture implementation plus values of the neural net weights. In these examples, a trained learning model takes data as an input, and produces an output such as a cluster-set, a prediction, a class-label, a regression value, a recommended action, or the like, as output (other types of responses/outcomes are possible as well).” Examiner Note: When Stöcker’s node and network are applied to Calmon’s decentralized machine learning, the resulting system would include a master node obtaining Calmon’s parameterized learning model and training model, wherein the shared training parameters are based on a local model of the participant nodes being trained during the most recent iteration.);
generate merged training parameters based on the shared training parameters ([0053] “Meanwhile, FIG. 4B provides an example of a process 400B in which both LSP nodes 410 and 420 are improving a common learning model incrementally over time. In one embodiment this is done by using a neural network 440 at an LSP (which in the example of FIG. 4B is the LSP node 410). The neural network 440 may use outputs generated by the existing model and an incremental model (trained for example on a new bucket of data) during training. In some embodiments, for a given input feature vector it may compute a mathematical combination of its output, the output from the existing model and the output from the incremental model to compute an error value used to train its weights. Once training is complete, the neural network 440 becomes the updated model that will be communicated by the LSP to the verification node and/or other LSP.” Examiner Note: When Stöcker’s node and network are applied to Calmon’s decentralized machine learning, the resulting system would generate merged training parameters based on the shared training parameters.);
generate a transaction that includes an indication that the master node has generated the merged training parameters ([0053] “Once training is complete, the neural network 440 becomes the updated model that will be communicated by the LSP to the verification node and/or other LSP.” Examiner Note: When Stöcker’s node and network are applied to Calmon’s decentralized machine learning, the resulting system would generate a transaction that includes an indication that the master node has generated the merged training parameters);
cause the transaction to be written as a block on the distributed ledger ([0055] “For example, the verification node can provide the learning service provider with an encryption key for decrypting the test data after the learning service provider delivers a learning model to the verification node or stores it on the blockchain.” Examiner Note: When Stöcker’s node and network are applied to Calmon’s decentralized machine learning, the resulting system would cause the ML transaction to be written as a block on the distributed ledger).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 9, the combination of Stöcker, Kate and Calmon as detailed above teaches The system of claim 8.
Stöcker teaches wherein the self-healing computer node is further programmed to: based on the triggering of the corrective action, obtain the parameters from the master node ([0146] “Preferably, directly upon receipt of the data synchronization indication the communication module 430.2 may start/trigger one or more data synchronization action(s) (step 804). As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400. Preferably, the communication module 430.2 transmits said message to the directly neighbored node 404.1 via the (shortest) communication connection 428.” [0146] “Preferably, directly upon receipt of the data synchronization indication the communication module 430.2 may start/trigger one or more data synchronization action(s) (step 804). As a first data synchronization action, the communication module 430.2 may be configured to transmit a data synchronization request message to at least one further node 402, 404.1 of the peer-to-peer network 400. Preferably, the communication module 430.2 transmits said message to the directly neighbored node 404.1 via the (shortest) communication connection 428.”);
apply the parameters to the update the local state at the self- healing computer node ([0151] “Upon receipt of the data packet(s) from the further node 404.1 the requesting node 404.2 may update its data of the peer-to-peer application (step 807).”).
Stöcker teaches wherein the self-healing computer node is further programmed to: based on the triggering of the corrective action, obtain the parameters from the master node; apply the parameters to the update the local state at the self- healing computer node, but does not explicitly teach merged training parameters; the local ML state.
Calmon teaches merged training parameters ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” [0053] “Meanwhile, FIG. 4B provides an example of a process 400B in which both LSP nodes 410 and 420 are improving a common learning model incrementally over time. In one embodiment this is done by using a neural network 440 at an LSP (which in the example of FIG. 4B is the LSP node 410). The neural network 440 may use outputs generated by the existing model and an incremental model (trained for example on a new bucket of data) during training. In some embodiments, for a given input feature vector it may compute a mathematical combination of its output, the output from the existing model and the output from the incremental model to compute an error value used to train its weights. Once training is complete, the neural network 440 becomes the updated model that will be communicated by the LSP to the verification node and/or other LSP.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically merges ML training parameters from the master node.);
the local ML state ([0021] “Example embodiments provide methods, devices, networks and/or systems, which support a blockchain network capable of implementing a mutually exchanged and trusted learning environment. In some embodiments, one entity may learn based on learning capabilities of another entity, while in other embodiments, multiple entities may learn from each other through a mutual exchange of learning models and data.” [0038] “In a second embodiment, there are multiple participants each of which is a data owner and model builder, and some additional verification nodes, and there is a single learning model that the data owners/model builders progressively update over multiple iterations. In this case, after the verification nodes have generated the test/train sets, an iterative process of model sharing may occur where each data-owner/model-builder participant takes turns to improve the model and send it to the verification nodes.” Examiner Note: When Stöcker’s node and network is applied to Calmon’s decentralized ML system including training a machine-learned model, the resulting system specifically synchronizes the node with ML training parameters and an ML state.).
Stöcker, Kate, and Calmon are analogous art because they are directed towards cryptographic networking. Therefore, it would have been obvious to one of ordinary skill in the art prior to the effective filing date of the claimed invention to combine Stöcker and Kate’s network with Calmon’s decentralized machine learning. The combination would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention because they would have been motivated to increase the flexibility and user base of the system, which can be accomplished by adapting the system to blockchain based machine learning (Calmon [0022] “Some of the benefits of such a system include the ability of a data owner to receive benefit to their proprietary data through mutual exchange of data and learning models with another similar data owner. In this way, each learning service provider can use their own proprietary data to improve the learning of another party without having to divulge their proprietary data. Blockchain is different from a traditional database in that blockchain is not a central storage but rather a decentralized, immutable, and secure storage, where nodes must share in changes to records in the storage. Some properties that are inherent in blockchain and which help implement the blockchain include, but are not limited to, an immutable ledger, smart contracts, security, privacy, decentralization, consensus, endorsement, accessibility, and the like, which are further described herein. According to various aspects, the mutual learning is implemented in a trusted and fair manner because of the immutable ledger and the distributed network which are inherent and unique to blockchain. In particular, verification nodes and the blockchain can provide a layer of verification between a data provider and a learning service provider to ensure that the learning service provider trains a generalizable learning model (i.e., a model that will perform well on new, previously unseen data) and does not just create a seemingly high-performing (but actually non-generalizing) model based on knowing the data that the model's performance is going to be tested and measured against. Furthermore, the verification nodes may ensure that the data provider satisfies their obligations such as payment or even another learning model enhancing the learning service provider's data before receiving access to the learning model created by the learning service provider.”).
As per claim 10, the combination of Stöcker, Kate and Calmon as detailed above teaches The system of claim 1.
Stöcker teaches wherein the fault condition within the blockchain network is a network connectivity outage, power outage, or computing node crash ([0143] “In a first step 801, e.g. node 404.2 may detect, by using the detecting means 458.2, that an operation error of the node 404.2 has occurred. By way of example, the detecting means 458.2 detect a reboot process of the node 404.2. Such a reboot process may be an (indirect) indication of a previously occurred operation error, such as a power outage or a communication connection interruption with the other nodes 402, 404.1 of the peer-to-peer network 400.”).
As per claim 11, the combination of Stöcker, Kate and Calmon as detailed above teaches The system of claim 9.
Stöcker teaches wherein the self-healing computer node is further programmed to: automatically perform a node restart to recover from the fault condition ([0143] “In a first step 801, e.g. node 404.2 may detect, by using the detecting means 458.2, that an operation error of the node 404.2 has occurred. By way of example, the detecting means 458.2 detect a reboot process of the node 404.2. Such a reboot process may be an (indirect) indication of a previously occurred operation error, such as a power outage or a communication connection interruption with the other nodes 402, 404.1 of the peer-to-peer network 400.”).
Claim 12 is a method claim corresponding to system claim 1. Claim 12 is rejected for the same reasons as claim 1.
Claim 13 is a method claim corresponding to system claim 3. Claim 13 is rejected for the same reasons as claim 3.
Claim 14 is a method claim corresponding to system claim 4. Claim 14 is rejected for the same reasons as claim 4.
Claim 15 is a method claim corresponding to system claim 5. Claim 15 is rejected for the same reasons as claim 5.
Claim 16 is a method claim corresponding to system claim 6. Claim 16 is rejected for the same reasons as claim 6.
Claim 17 is a method claim corresponding to system claim 7. Claim 17 is rejected for the same reasons as claim 7.
Claim 18 is a method claim corresponding to system claim 9. Claim 18 is rejected for the same reasons as claim 9.
Claim 19 is a method claim corresponding to system claim 11. Claim 19 is rejected for the same reasons as claim 11.
Claim 20 is a non-transitory machine-readable storage medium analogous to the system of claim 1. Claim 20 requires A non-transitory machine-readable storage medium comprising instructions executable by a processor of a self-healing computer node of a blockchain network comprising a plurality of computing nodes, the self-healing computer node recovering from a fault condition within the blockchain network, the instructions programming the processor (Stocker, [0011] “…A peer-to-peer network, also called computer-to-computer network, comprises a plurality of [computer] nodes. In particular, at least one first node and at least one further or second node can be provided. Each of these nodes comprises at least a part of a peer-to-peer application. In the peer-to-peer application data can be stored and/or processed…”). Claim 20 is rejected for the same reasons as claim 1.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LUIS A SITIRICHE whose telephone number is (571)270-1316. The examiner can normally be reached M-F 9am-6pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, David Yi can be reached at (571) 270-7519. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/LUIS A SITIRICHE/Primary Examiner, Art Unit 2126
Prosecution Insights