DETAILED ACTION
This office action is in response to amendments filed on 04/21/2026.
Claims 1-2, 4-7, 9-12, 14-17, and 19-20 have been amended. Claims 3 and 13 have been canceled. Claims 1-2, 4-12, and 14-20 are pending.
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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 04/30/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Arguments
Rejections Under 35 USC § 112(b):
In light of applicant’s amendments to the claims (pg. 2-6), the rejections under 35 USC § 112(b) have been withdrawn.
Rejections Under 35 USC § 101:
In light of applicant’s amendments to the claims (pg. 2-6) and arguments (pg. 8-9), the rejections under 35 USC § 101 have been withdrawn.
Rejections Under 35 USC § 103:
Applicant's arguments regarding the prior art rejections have been fully considered but they are not persuasive.
Applicant argues (pg. 10) that the authentication tokens disclosed by Khaleghi are used for data access control rather than encrypted machine learning model training. Examiner respectfully notes that Khaleghi is not relied upon for this purpose. As can be seen in the rejection below, Badawi teaches the use of public and private keys for encrypted machine learning model training. Khaleghi is simply relied upon for generation of those keys based on user information.
Applicant argues (pg. 10-11) that Khaleghi teaches generation of a single key or authentication token, while the claims require the generation of a separate public/private key pair. Examiner respectfully notes that Khaleghi is not relied upon for this purpose. As can be seen in the rejection below, Badawi teaches the separate public/private key pair. Khaleghi is simply relied upon for generation of those keys based on user information.
Applicant argues (pg. 11) that Badawi does not teach encryption using a cryptographic hash function as recited in the amended independent claim. Examiner respectfully notes that, as can be seen in the rejection below, Khaleghi teaches encryption using a cryptographic hash function. Swapping Badawi’s CKKS encryption scheme for Khaleghi’s cryptographic hash function amounts to a simple substitution of known alternative encryption schemes, and thus would have been obvious to one of ordinary skill in the art.
Applicant argues (pg. 11-12) that Badawi’s model does not depend on provision of the user-specific keys at prediction time, and that there is no disclosure that the model would generate no prediction or an inaccurate prediction in the absence of the keys. Examiner respectfully disagrees. As explained in the rejection below, in Badawi, in order to generate the model output prediction, the
s
o
f
t
m
a
x
function must be applied to the decrypted ciphertext. Decryption can only be performed using the intended party’s personal private key. Therefore, if the intended party’s model is accessed by another non-intended party without providing the private key, the model will not generate an output prediction.
The prior art rejections have been updated to include the amended limitations and to clarify the reasoning given for the limitations that were not amended.
Claim Rejections - 35 USC § 103
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.
Claims 1-4, 6, 11-14, and 16 are rejected under 35 U.S.C. 103 as being unpatentable over
Badawi et al. (hereinafter Badawi), “PrivFT: Private and Fast Text Classification with Homomorphic Encryption” in view of
Khaleghi, Untied States Patent US 10735191 B1.
Regarding Claim 1,
Badawi teaches A computer-implemented method of training a machine learning model, comprising: (Pg. 9, section 5.2: “PrivFT is implemented in two libraries on two different hardware platforms. The first implementation uses Microsoft SEAL v3.3.0 and runs on CPU. The second implementation utilizes our GPU implementation of the CKKS scheme - described in the previous section and runs on NVIDIA-enabled GPUs.”)
Generating, in response to the verification, a private key and a public key for the user [based on the user information]; (Pg. 5, section 3.4: “Given input data represented as real or complex numbers, we use a modified version of CKKS [14] for encryption
(
E
N
C
)
and decryption
(
D
E
C
)
as follows…
K
E
Y
G
E
N
: generate: (1) secret key
s
←
X
k
e
y
∈
R
q
L
, and (2) public key
(
a
,
b
)
∈
R
q
L
2
, where
a
←
X
q
L
and
b
=
-
a
s
+
e
with
e
←
X
e
r
r
.” A secret key is a private key.)
receiving input bytes containing user-specific features; (Pg. 2, section 1.1: “To give motivational use-cases for PrivFT, consider the inference and training as a service shown in Figure 1… In Figure 1b, Alice has a private dataset and wants to train a model on the cloud.” The dataset used for training (i.e. input bytes) is private to a user (i.e. contains user-specific features).)
converting the input bytes into encrypted bytes by applying a [cryptographic hash function] to the input bytes combined with the private key and the public key; (Pg. 2, section 1.1: “In Figure 1b, Alice has a private dataset and wants to train a model on the cloud. She encrypts her dataset and sends to the cloud. The cloud runs the training algorithm and generates an encrypted model that is communicated back to Alice.” Pg. 5, section 3.4: “Given input data represented as real or complex numbers, we use a modified version of CKKS [14] for encryption
(
E
N
C
)
and decryption
(
D
E
C
)
as follows…
K
E
Y
G
E
N
: generate: (1) secret key
s
←
X
k
e
y
∈
R
q
L
, and (2) public key
(
a
,
b
)
∈
R
q
L
2
, where
a
←
X
q
L
and
b
=
-
a
s
+
e
with
e
←
X
e
r
r
…
E
N
C
(
µ
)
: given a plaintext message
(
µ
)
, sample
u
←
X
q
L
and
e
0
,
e
1
←
X
e
r
r
. Return ciphertext
c
t
=
(
c
0
,
c
1
)
=
(
a
v
+
µ
+
e
0
,
b
v
+
e
1
)
∈
R
q
L
2
.
D
E
C
(
c
t
)
: given a ciphertext
c
t
∈
R
q
l
2
, return
µ
=
c
0
+
s
c
1
∈
R
q
l
.” The training dataset (i.e. input bytes) is encrypted (i.e. converted to encrypted bytes) prior to being sent to the cloud for training. Encryption
E
N
C
returns ciphertext computed based on the public key
(
a
,
b
)
, which can be decrypted using the secret (private) key
s
(i.e. the encryption is based on the private and public keys).)
feeding the encrypted bytes, the private key, and the public key into a machine learning model; (Pg. 2, section 1.1: “She encrypts her dataset and sends to the cloud. The cloud runs the training algorithm and generates an encrypted model that is communicated back to Alice.” The encrypted training dataset (i.e. encrypted bytes), which is encrypted based on the private and public keys, is sent to the cloud for model training (i.e. fed into a machine learning model).)
training the machine learning model based on the encrypted bytes, the private key, and the public key, (Pg. 2, section 1.1: “We demonstrate how to train an effective model using an encrypted dataset with FHE [Fully Homomorphic Encryption].” The machine learning model is trained using the encrypted training dataset (i.e. encrypted bytes), which is encrypted based on the private and public keys.)
wherein the training causes parameters of the machine learning model to be optimized for the user; and (Pg. 2, section 1.1: “In Figure 1b, Alice has a private dataset and wants to train a model on the cloud. She encrypts her dataset and sends to the cloud. The cloud runs the training algorithm and generates an encrypted model that is communicated back to Alice.” Pg. 7, section 4.2: “The training procedure is used to find the weights of the hidden and output layers
H
and
O
.” The model is trained to find the optimal weights of the hidden and output layers (i.e. optimize parameters of the machine learning model). Since the model is trained on the user’s private dataset, its learned parameters will be optimized for the user.)
generating a personalized machine learning model for the user based on the training of the machine learning model, (Pg. 2, section 1.1: “The cloud runs the training algorithm and generates an encrypted model that is communicated back to Alice. Alice can decrypt the model and use it for local inference.” The training algorithm generates a model based on the user’s data (i.e. a personalized model) and sends it to the user for decryption and use by the user.)
wherein the personalized machine learning model is configured to generate either no prediction or an inaccurate prediction when the private key or the public key are not provided. (Pg. 1, section 1: “The key assumption here is that as long as the data is encrypted with a secure cryptographic scheme using a private key that only intended parties have, encrypted data can be released in public. FHE encrypted data can be operated on by an untrusted evaluator to generate encrypted results without revealing intermediate/final results to the evaluator. The results are communicated back to the data owner who can decrypt and make use of them.” Pg. 7, section 4.1: “To compute the class scores
s
… The resultant ciphertexts are summed to generate a packed ciphertext encrypting
s
, which is communicated back to the client who can decrypt and find the best class by evaluating the
s
o
f
t
m
a
x
function in plaintext.” In order to generate the model output, the
s
o
f
t
m
a
x
function must be applied to the decrypted ciphertext. Decryption can only be performed using the intended party’s private key. Therefore, if the intended party’s model is accessed by another non-intended party without providing the private key, the model will not generate an output.)
Badawi does not appear to explicitly disclose
receiving user information for a user;
verifying the user information against a permissions database, wherein verifying comprises confirming that the user information corresponds to an authorized user;
generating keys based on the user information;
encryption using a cryptographic hash function
However, Khaleghi teaches receiving user information for a user; (Col. 15, lines 46-49: “At block 602, the user device may access local user authentication data (e.g., a password received from the user, biometric data of the user, and/or the like).” User authentication data is user information.)
verifying the user information against a permissions database, wherein verifying comprises confirming that the user information corresponds to an authorized user; (Col. 12, lines 30-43: “FIG. 3B illustrates an example authenticator subsystem that may be included in the applications 302 (e.g., in the data sharing application 304), that may be utilized to authenticate a user and to determine what data and application rights the user is to be granted. As illustrated, the authenticator subsystem may include a facial recognition module 302B configured to recognize a user based on an image captured with a user device camera. Features extracted from the image may be compared to stored features of an authorized user. If there is a match, the user may be granted with corresponding rights (e.g., access may be granted to certain application functions and/or user information). If the facial features fail to match, the user may not be granted such rights.” Facial recognition data (i.e. user information) is compared to stored features of an authorized user (i.e. against a permissions database) to determine whether the user corresponds to an authorized user.)
generating keys based on the user information; (Col. 15, lines 49-55: “At block 604, a key is generated using the authentication data. At block 606, an authentication token is generated. The authentication token may be different than the authentication data and different than the key. For example the authentication [token] may be generated using a key derivation function, such as Password-Based Key Derivation Function…” A cryptographic key is generated based on the authentication data (i.e. user information).)
encryption using a cryptographic hash function (Col. 5, lines 59-63: “Optionally, encryption may be performed using symmetric keys, or public/private key pairs. The data may be encrypted using a cryptographic hash, such as, by way of example, MD5, SHA1, SHA256, SHA384, or SHA512 hashes.”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Badawi and Khaleghi. Badawi teaches homomorphically encrypted machine learning inference and training for text classification. Khaleghi teaches controlling access to sensitive data using cryptographic keys generated using a password-based key derivation function. One of ordinary skill would have motivation to combine Badawi and Khaleghi in order to “make it extremely difficult for an adverse party (e.g., a hacker) to determine the authentication token using brute force guessing. For example, a 90 bits password processed using 4500 PBKDF2 [Password-Based Key Derivation Function 2], may take trillions of years to guess, even with a high powered graphical processing unit,” (Khaleghi, col. 15, lines 58-63) thereby “enable[ing] communication of data and information, including multimedia data, among disparate systems in a secure and controlled manner” (Khaleghi, col. 17, lines 6-8).
Regarding Claim 2, Badawi and Khaleghi teach The computer-implemented method of claim 1, as shown above.
Khaleghi also teaches wherein the user information comprises a user identification and a password associated with an application. (Col. 6, lines 63-67: “[T]he user may be authenticated a first time using a username and a password (e.g., within a session of limited duration), a time-limited token may be generated and provided in return by the protected resource, and the token may for further authentication…” A username is a user identification.)
Regarding Claim 4, Badawi and Khaleghi teach The computer-implemented method of claim 1, as shown above.
Badawi also teaches wherein the encrypted bytes are provided as inputs to a neural embedding matrix to generate an embedding for use as input to the machine learning model during training. (Pg. 6, section 4.1: “we require the client to encode her text
r
1
,
.
.
.
,
r
w
into a 1-hot vector
(
v
)
… The client encrypts
v
and sends it along with the number of words
w
(in plaintext) to the server. By doing so, the server can compute the embedded representation
h
(Step 2) by a vector-matrix multiplication
(
v
·
H
)
and a plaintext multiplication
(
H
M
U
L
P
L
A
I
N
)
of the factor
1
/
w
. We assume here that the embedding vectors in
H
correspond to the words in the same order as indexed in the dictionary.” Matrix
H
is a neural embedding matrix, and the encrypted input
v
(i.e. encrypted bytes) is provided to the neural embedding matrix to generate the embedded representation
h
which is input to the model.)
Regarding Claim 6, Badawi and Khaleghi teach The computer-implemented method of claim 1, as shown above.
Badawi also teaches wherein training the machine learning model comprises optimizing parameters of the machine learning model via backpropagation using labeled training data that has been converted into the encrypted bytes to reflect the user-specific features. (Pg. 2, section 1.1: “Here, the data owner sends her encrypted data, using an FHE scheme, to the cloud which in turn performs a batched training algorithm using back propagation to learn an encrypted model (i.e. training operations are done entirely in ciphertext space)… In Figure 1b, Alice has a private dataset and wants to train a model on the cloud. She encrypts her dataset and sends to the cloud. The cloud runs the training algorithm and generates an encrypted model that is communicated back to Alice.” Pg. 7, section 4.2: “The training procedure is used to find the weights of the hidden and output layers
H
and
O
.” Pg. 9, section 5.2.2: “[W]e had to adapt the training procedure in fasttext to use minibatch SGD… calculating the error function between each of the scores in the output vector and the truth label.” The model is trained via backpropagation using encrypted training data with truth labels to find the optimal weights of the hidden and output layers (i.e. optimize parameters of the machine learning model). Since the model is trained on the user’s private dataset, its learned parameters will reflect the user-specific features.)
Claims 11-14 and 16 are system claims containing substantially the same elements as method claims 1-4 and 6, respectively. Badawi and Khaleghi teach the elements of claims 1-4 and 6, as shown above.
Badawi also teaches A system for training a machine learning model, comprising: a processor; and a memory, coupled to the processor, configured to store executable instructions that, when executed by the processor, cause the processor to perform operations including: (Examiner notes that this limitation is interpreted as implementation of the disclosed method in a generic computing environment. Pg. 9, section 5.2: “PrivFT is implemented in two libraries on two different hardware platforms. The first implementation uses Microsoft SEAL v3.3.0 and runs on CPU. The second implementation utilizes our GPU implementation of the CKKS scheme - described in the previous section and runs on NVIDIA-enabled GPUs.”)
Claims 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Badawi in view of Khaleghi and further in view of
Yao et al. (hereinafter Yao), “Dynamic Word Embeddings for Evolving Semantic Discovery”.
Regarding Claim 5, Badawi and Khaleghi teach The computer-implemented method of claim 1, as shown above.
Badawi also teaches wherein training the machine learning model comprises applying [dynamic] embedding of the encrypted bytes to generate encrypted neural embeddings and performing one or more layers of linear or nonlinear neural applications using the encrypted neural embeddings as inputs. ((Pg. 6, section 4.1: “we require the client to encode her text
r
1
,
.
.
.
,
r
w
into a 1-hot vector
(
v
)
… The client encrypts
v
and sends it along with the number of words
w
(in plaintext) to the server. By doing so, the server can compute the embedded representation
h
(Step 2) by a vector-matrix multiplication
(
v
·
H
)
and a plaintext multiplication
(
H
M
U
L
P
L
A
I
N
)
of the factor
1
/
w
. We assume here that the embedding vectors in
H
correspond to the words in the same order as indexed in the dictionary… To compute the class scores
s
, a slightly different approach is used to multiply
h
with the output matrix
O
. Since
h
is non-packed, we multiply each component
h
i
by the rows of
O
to generate
n
packed ciphertexts. This step requires
n
H
M
U
L
P
L
A
I
N
operations. The resultant ciphertexts are summed to generate a packed ciphertext encrypting
s
, which is communicated back to the client who can decrypt and find the best class by evaluating the
s
o
f
t
m
a
x
function in plaintext.” Multiplying the encrypted input
v
by the embedding matrix
H
to derive the embedded representation
h
amounts to embedding of the encrypted bytes to generate encrypted neural embeddings, and multiplying the embedded representation
h
by the output layer matrix
O
amounts to performing a layer of linear neural applications using the encrypted neural embeddings as inputs.)
Badawi and Khaleghi do not appear to explicitly disclose applying dynamic embedding
However, Yao teaches applying dynamic embedding (Pg. 1, abstract: “In this paper, we develop a dynamic statistical model to learn time-aware word vector representation. We propose a model that simultaneously learns time-aware embeddings and solves the resulting ‘alignment problem’.”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Badawi, Khaleghi, and Yao. Badawi teaches homomorphically encrypted machine learning inference and training for text classification. Khaleghi teaches controlling access to sensitive data using cryptographic keys generated using a password-based key derivation function. Yao teaches learning dynamic, time-aware word embeddings to model the evolution of word associations and meanings over time. One of ordinary skill would have motivation to combine Badawi, Khaleghi, and Yao because “understanding and tracking word evolution is useful for time-aware knowledge extraction tasks (e.g., public sentiment analysis), and other applications in text mining” (Yao, pg. 1, section 1). “Our qualitative and quantitative tests indicate that our method not only reliably captures this evolution over time, but also consistently outperforms state-of-the-art temporal embedding approaches on both semantic accuracy and alignment quality” (Yao, pg. 1, abstract).
Claim 15 is a system claim containing substantially the same elements as method claim 5. Badawi, Khaleghi, and Yao teach the elements of claim 5, as shown above.
Claims 7-9 and 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Badawi in view of Khaleghi and further in view of
Atrey et al. (hereinafter Atrey), “Preserving Privacy in Personalized Models for Distributed Mobile Services”.
Regarding Claim 7, Badawi and Khaleghi teach The computer-implemented method of claim 1, as shown above.
Badawi and Khaleghi imply but do not appear to explicitly disclose the remaining features of claim 7.
However, Atrey teaches wherein generating a personalized machine learning model for the user comprises obtaining a first personalized machine learning model for a first user by training with a first private key and a first public key generated from first user information and obtaining a second personalized machine learning model for a second user by training with a second private key and a second public key generated from second user information, wherein the first private key and the first public key are different from the second private key and the second public key. (Pg. 7, section V.A: “Once a general ML model has been trained in the cloud, the next phase personalizes this model for each user using transfer learning. The personalization involves using a small amount of training data for each new user to learn a distinct personalized model,
M
P
.”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Badawi, Khaleghi, and Atrey. Badawi teaches homomorphically encrypted machine learning inference and training for text classification. Khaleghi teaches controlling access to sensitive data using cryptographic keys generated using a password-based key derivation function. Atrey teaches generating personalized models for each user of a service while preserving data privacy. One of ordinary skill would have motivation to combine Badawi, Khaleghi, and Atrey because “[p]ersonalized models can encode specific behavior exhibited by an individual user and offer better efficacy over an aggregated model” (Atrey, pg. 2, section II). However, “personalized ML models encode sensitive information in the single-user context traces used as training data and mobile services that use such personalized models can leak privacy information through a class of privacy attacks known as model inversion”, and Atrey’s Pelican framework can “reduce privacy leakage up to 75%” (Atrey, pg. 1-2, section I).
Regarding Claim 8, Badawi, Khaleghi, and Atrey teach The computer-implemented method of claim 7, as shown above.
Atrey also teaches wherein the first personalized machine learning model includes a first set of model parameters optimized for the first user and the second personalized machine learning model includes a second set of model parameters optimized for the second user. (Pg. 8, section V.A: “In our case, as new personal data becomes available, the transfer learning process can be re-invoked to update the parameters of the personalized model, after which it is redeployed for use by the service.” The parameters of each user’s personalized model are updated based on personal data for the user (i.e. the model parameters are optimized for the user).)
Regarding Claim 9, Badawi, Khaleghi, and Atrey teach The computer-implemented method of claim 7, as shown above.
Badawi also teaches wherein, when the first user accesses the second personalized machine learning model by providing the first private key and the first public key without providing user information associated with the second user, the second personalized machine learning model generates an output with an accuracy below a threshold or does not generate an output. (Pg. 1, section 1: “The key assumption here is that as long as the data is encrypted with a secure cryptographic scheme using a private key that only intended parties have, encrypted data can be released in public. FHE encrypted data can be operated on by an untrusted evaluator to generate encrypted results without revealing intermediate/final results to the evaluator. The results are communicated back to the data owner who can decrypt and make use of them.” Pg. 7, section 4.1: “To compute the class scores
s
… The resultant ciphertexts are summed to generate a packed ciphertext encrypting
s
, which is communicated back to the client who can decrypt and find the best class by evaluating the
s
o
f
t
m
a
x
function in plaintext.” In order to generate the model output, the
s
o
f
t
m
a
x
function must be applied to the decrypted ciphertext. Decryption can only be performed using the intended party’s private key. Therefore, if the intended party’s (i.e. second user’s) model is accessed by another non-intended party (i.e. first user) by providing the non-intended party’s keys (i.e. first private key and first public key) without providing the intended party’s private key, the model will not generate an output.)
Claims 17-19 are system claims containing substantially the same elements as method claims 7-9, respectively. Badawi, Khaleghi, and Atrey teach the elements of claims 7-9, as shown above.
Claims 10 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Badawi in view of Khaleghi and further in view of
Yang et al. (hereinafter Yang), “NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications”.
Regarding Claim 10, Badawi and Khaleghi teach The computer-implemented method of claim 1, as shown above.
Badawi also teaches wherein training the machine learning model is based on the encrypted bytes, the private key, the public key, [and the device information of the device], (Pg. 2, section 1.1: “We demonstrate how to train an effective model using an encrypted dataset with FHE [Fully Homomorphic Encryption].” The machine learning model is trained using the encrypted training dataset (i.e. encrypted bytes), which is encrypted based on the private and public keys.)
Badawi and Khaleghi do not appear to explicitly disclose the remaining features of claim 10.
However, Yang teaches wherein prior to training the machine learning model, the method further comprises receiving device information of a device intended to run the personalized machine learning model for the user, wherein the device information comprises at least one of a device constraint type indicating a runtime requirement, a memory size requirement, or a quality requirement, (Pg. 2, figure 1: “NetAdapt automatically adapts a pretrained network to a mobile platform given a resource budget. This algorithm is guided by the direct metrics for resource consumption.” Pg. 4, section 3.1: “The resource can be latency, energy, memory footprint, etc., or a combination of these metrics.” The mobile platform is a device intended to run the machine learning model, and a resource budget is device information. The resource budget (i.e. device information) can be, for example, a latency budget (i.e. a runtime requirement) or a memory footprint budget (i.e. a memory size requirement).)
wherein training the machine learning model is based on […] the device information of the device, and wherein the personalized machine learning model has a model architecture and complexity selected based on the device constraint type. (Pg. 6, figure 2: “At each iteration, NetAdapt decreases the resource consumption by simplifying (i.e., removing filters from) one layer. In order to maximize accuracy, it tries to simplify each layer individually and picks the simplified network that has the highest accuracy. Once the target budget is met, the chosen network is then fine-tuned again until convergence.” The layers of the model are simplified by removing filters (i.e. the model architecture and complexity is selected) and the model is fine-tuned (i.e. trained) in order to meet the resource budget constraint of the target device (i.e. based on the device information/device constraint type).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Badawi, Khaleghi, and Yang. Badawi teaches homomorphically encrypted machine learning inference and training for text classification. Khaleghi teaches controlling access to sensitive data using cryptographic keys generated using a password-based key derivation function. Yang teaches adapting a neural network to a target device based on a resource budget constraint of a target device. One of ordinary skill would have motivation to combine Badawi, Khaleghi, and Yang because “DNN-based AI applications are typically too computationally intensive to be deployed on resource-constrained platforms, such as mobile phones. This hinders the enrichment of a large set of user experiences” (Yang, Pg. 1, section 1). Yang’s algorithm “outperforms the state-of-the-art automatic network simplification algorithms by up to
1.7
×
in terms of reduction in measured inference latency while delivering equal or higher accuracy” (Yang, Pg. 3, section 1).
Claim 20 is a system claim containing substantially the same elements as method claim 10. Badawi, Khaleghi, and Yang teach the elements of claim 10, as shown above.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BENJAMIN M ROHD whose telephone number is (571)272-6445. The examiner can normally be reached Mon-Thurs 8:00-6:00 EST.
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, Viker Lamardo can be reached at (571) 270-5871. 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.
/B.M.R./Examiner, Art Unit 2147
/VIKER A LAMARDO/Supervisory Patent Examiner, Art Unit 2147
Prosecution Insights