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 .
Response to Arguments
Applicant claims that the 112b rejections regarding claims 5-7 should be removed, as their meaning is “clear to the skilled person (Remarks, pg. 11).” While the claim uses “arranged in a matrix” and “positions (x−1,y)…(x−3,y),” which suggests indices, the claims are still problematic because x and y are never explicitly defined as matrix indices; they are used as free variables and the reader has to infer their domain and role. Further, “a status identifier s_{x,y} for a position (x,y) in the matrix” strongly implies indices, but still lacks an express statement that x and y are indices (e.g., row/column integers).
Applicant argues (Remarks, pg. 12) that no abstract ideas are recited by the claims, as there exists an example of a claim in the MPEP that does not recite an abstract idea that does recite a neural network. However, the claims at-hand are not analogous to the example provided by the MPEP.
Applicant argues that the claims cannot be analyzed under 35 U.S.C. 101 as they cannot be practically performed in the human mind (Remarks, pg. 12). However, “Data compression” and “decoding” are, at the level claimed, data manipulation / mathematical processing. Applicant characterizes “context-dependent arithmetic coding” as a “universal compression technique,” which reinforces that the claim is directed to an algorithmic encoding/decoding scheme, i.e., a mathematical concept and/or mental process in the Step 2A Prong One sense when claimed as operations on symbols/bits/probabilities without a specific technological implementation that changes the character of the claim. See MPEP § 2106.04(a) and § 2106.04(a)(2). Even accepting that a human is not practically suited to perform modern compression at scale, that does not remove the claim from the mental-process/mathematical-concept categories when the claim recites the abstract operations themselves (decoding/deriving/assigning numerical parameters) rather than a specific technological improvement. MPEP § 2106 and § 2106.04(a)(2).
Applicant’s argument that Golomb-Rice coding is ‘fundamentally different’ from context-dependent arithmetic coding (e.g., CABAC) is not persuasive (Remarks, pg. 16-18). The rejection does not allege equivalence; rather, the rejection relies on the applied references collectively to teach the claimed ‘context-dependent arithmetic coding’ and associated context selection/decoding features. In particular, unlike CABAC (Context-Based Adaptive Binary Arithmetic Coding), which is a specific standardized algorithm, "context-dependent arithmetic coding" is an ad-hoc descriptive phrase subject to BRI. A true term of art must have a consistent meaning to a Person Having Ordinary Skill in the Art (PHOSITA). However, "context" is defined radically differently across competing standards, and because the definition of "context" changes based on the application rather than the term itself, it cannot be a universal term of art. This results in a "zone of uncertainty" that makes it impossible for the PHOSITA to know what is being claimed.
Applicant’s assertion that Marpe is ‘silent’ regarding neural-network weights is also not persuasive because the pertinent teaching of Marpe is the context-dependent arithmetic coding technique itself, which is applied to symbol/bit streams irrespective of the semantic origin of the data, and a reference need not be limited to the same end-use environment to be combinable (Remarks, pg. 17). See MPEP §§ 2141, 2143; § 2141.01(a).
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
Claims 5-7 are rejected under 35 U.S.C. §112(b) as being indefinite. The variables “x” and “y” are
used to define positions or locations, but these terms are not introduced or defined in the claims. The
absence of a description for these variables renders it unclear what they refer to, how they are
determined, or whether they represent matrix indices, coordinates, or other structures. As a result, the
metes and bounds of the claimed invention cannot be ascertained.
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.
Claims 1-17 are rejected under 35 U.S.C. 101 because they are directed to an abstract
idea without significantly more.
Regarding claim 1 and analogous claims 13 and 14:
Step 1: is the claim directed to one of the four statutory categories?
Yes, the claim is directed to a machine.
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract idea?
Yes. The limitations: “decoding the weight parameters of the neural network using a context-dependent arithmetic coding; selecting a context for a decoding of a weight parameter, or for a decoding of a syntax element of a number representation of the weight parameter, in dependence on one or more previously decoded weight parameters and/or in dependence on one or more previously decoded syntax elements of a number representation of one or more weight parameters; decoding the weight parameter, or a syntax element of the weight parameter, using the selected context; ”and “and selecting a context out of two possible contexts for a decoding of a bit at a given bit position of the unary code or of the truncated unary code in dependence on a sign of the currently decoded weight parameter” are directed to a mental process of judgment and evaluation under MPEP 2106.04(a)(2)(III).
Step 2A, prong 2: Do the additional elements integrate into a practical application?
No. The limitation: “acquiring a plurality of bits representing weight parameters of the neural network;” is directed to mere data gathering under MPEP 2106.05(g).
Further, the limitations: “A decoder for decoding weight parameters of a neural network, the decoder comprising: a communication interface; a memory storing instructions; and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising:” are directed to mere instructions to apply under MPEP 2106.05(f).
Step 2B: Does the claim recite additional elements that amount to significantly more than the
judicial exception?
No. The limitation: “acquiring a plurality of bits representing weight parameters of the neural network;” is directed to well-understood, routine, and conventional activity of “Receiving or
transmitting data over a network” under MPEP 2106.05(d).
Further, the limitations: “A decoder for decoding weight parameters of a neural network, the decoder comprising: a communication interface; a memory storing instructions; and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising:” are directed to generic computing components under MPEP 2106.05(f). The claim as a whole does not amount to significantly more than the judicial exception.
Regarding claim 2:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “selecting a context for a decoding of a zero flag of the weight parameter in dependence on a sign of a previously decoded weight parameter” is directed to a mental process of judgment under MPEP 2106.04(a)(2)(III).
Regarding claim 3:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “determining a plurality of status identifiers representing statuses of a plurality of weight parameters at a plurality of positions relative to a position of a currently decoded weight parameter in the form of a numeric value, and combining the status identifiers, in order to acquire a context index value representing a context of the currently decoded weight parameter” is directed to a mental process of judgment under MPEP 2106.04(a)(2)(III).
Regarding claim 4:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “wherein the operations further comprise: selecting the context for the decoding of the zero flag of the weight parameter in dependence on how many zero-valued weight parameters and/or unavailable weight parameters in a row are adjacent to the currently decoded weight parameter” is directed to a mental process of judgment under MPEP 2106.04(a)(2)(III).
Regarding claim 5:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “The decoder of The decoder of claim 1 wherein the plurality of weight parameters is arranged in a matrix, and the weight parameters are denoted as Ix.1,y, lx-2,y and lx-3,y and correspond to positions (x- 1,y), (x-2,y) and (x-3,y) in the matrix, respectively, and are represented by status identifiers sx-1,y, sx-2,y, sx-3,y” is directed to a mathematical concept under MPEP 2106.04(a)(2)(I).
Regarding claim 6:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. “The decoder of The decoder of claim 1wherein the plurality of weight parameters is arranged in a matrix, and a status identifier sx,y for a position (x,y) in the matrix is equal to a first value, if the position (x,y) is not available or the weight parameter at the position (x,y) is equal to zero, the status identifier sx,y for the position (x,y) is equal a second value, if the weight parameter at the position (x,y) is smaller than zero, and the status identifier sx,y for the position (x,y) is equal to a third value, if the weight parameter at the position (x,y) is larger than 0” is directed to a mathematical concept under MPEP 2106.04(a)(2)(I).
Regarding claim 7:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “The decoder of The decoder of claim 1 wherein the plurality of weight parameters is arranged in a matrix, and a status identifier sx,y for a position (x,y) in the matrix is equal to a first value, if the position (x,y) is not available or the weight parameter at the position (x,y) is equal to zero, and the status identifier sx,y for the position (x,y) is equal to a second value, if the position (x,y) is available and the weight parameter at the position (x,y) is not equal to zero” is directed to a mathematical concept under MPEP 2106.04(a)(2)(I).
Regarding claim 8:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “selecting a context for a decoding of a zero flag of the weight parameter in dependence on a distance of a closest non-zero weight parameter present in a predetermined direction, when seen from the currently decoded weight parameter” is directed to a mental process of judgment under MPEP 2106.04(a)(2)(III).
Regarding claim 9:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “selecting a context for a decoding of a zero flag of the weight parameter considering only a single one previously decoded weight parameter, which is adjacent to a currently decoded weight parameter” is directed to a mental process of judgment under MPEP 2106.04(a)(2)(III).
Regarding claim 10:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “herein the operations further comprises: determining a status identifier for the single one previously decoded weight position,wherein the status identifier for the single one previously decoded weight parameter equals to a first value, if the single one previously decoded weight parameter is not available or the weight parameter at the position (x,y) is equal to zero, equals to a second value, if the single one previously decoded weight parameter is smaller than zero, and equals to a third values, if the single one previously decoded weight parameter is larger than 0; andselecting the context in dependence on the status identifier” is directed to a mathematical concept under MPEP 2106.04(a)(2)(I).
Regarding claim 11:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “selecting a context associated with a zero value of the previously decoded weight parameter in case the previously decoded weight parameter is not available” is directed to a mental process of judgment under MPEP 2106.04(a)(2)(III).
Regarding claim 12:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “The decoder of The decoder of wherein the weight parameters are organized in rows and columns of a matrix, wherein an order in which the weight parameters are decoded is along a first row of the matrix, then along a subsequent second row of the matrix, or wherein an order in which the weight parameters are decoded is along a first column of the matrix, then along a subsequent second column of the matrix” are directed to a mathematical concept under MPEP 2106.04(a)(2)(I).
Regarding claim 15:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Step 1: is the claim directed to one of the four statutory categories?
Yes, the claim is directed to a machine.
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes. The limitation: “encoding the weight parameters of the neural network using a context-dependent arithmetic coding; selecting a context for an encoding of a weight parameter, or for an encoding of a syntax element of a number representation of the weight parameter, in dependence on one or more previously encoded weight parameters and/or in dependence on one or more previously encoded syntax elements of a number representation of one or more weight parameters; encoding the weight parameter, or a syntax element of the weight parameter, using the selected context;” and “and selecting a context out of two possible contexts for a decoding of a bit at a given bit position of the unary code or of the truncated unary code in dependence on a sign of the currently decoded weight parameter” is directed to a mental process of judgment and evaluation under MPEP 2106.04(a)(2)(III).
Step 2A, prong 2: Do the additional elements integrate into a practical application?
No. The limitation: “obtaining a plurality of weight parameters of the neural network;” is directed to mere data gathering under MPEP 2106.05(g).
Further, the limitations: “An encoder for encoding weight parameters of a neural network, the encoder comprising: a communication interface; a memory storing instructions; and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising:” are directed to mere instructions to apply under MPEP 2106.05(f).
Step 2B: Does the claim recite additional elements that amount to significantly more than the
judicial exception?
No. The limitation: “obtaining a plurality of weight parameters of the neural network;” is directed to well-understood, routine, and conventional activity of “Receiving or transmitting data over a network” under MPEP 2106.05(d).
Further, the limitations: “An encoder for encoding weight parameters of a neural network, the encoder comprising: a communication interface; a memory storing instructions; and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising” are directed to generic computing components under MPEP 2106.05(f). The claim as a whole does not amount to significantly more than the judicial exception.
Regarding claim 16:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes, the claim is dependent on claim 1.
Step 2A, prong 2: Do the additional elements integrate into a practical application?
No. The limitation: “A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using a decoder according to claim 1” is directed to field of use under MPEP 2106.05(h).
Step 2B: Does the claim recite additional elements that amount to significantly more than the
judicial exception?
No. The limitation: “A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using a decoder according to claim 1” is directed to field of use under MPEP 2106.05(h). The claim as a whole does not amount to significantly more than the judicial exception.
Regarding claim 17:
Step 2A, prong 1: Is the claim directed to a law of nature, a natural phenomenon, or an abstract
idea?
Yes, the claim is dependent on claim 1.
Step 2A, prong 2: Do the additional elements integrate into a practical application?
No. The limitation: “A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using an encoder according to claim 15” is directed to field of use under MPEP 2106.05(h).
Step 2B: Does the claim recite additional elements that amount to significantly more than the
judicial exception?
No. The limitation: “A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using an encoder according to claim 15” is directed to field of use under MPEP 2106.05(h). The claim as a whole does not amount to significantly more than the judicial exception.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1, 13-15 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by US Pre-Grant Patent 2018/0082181 (Brothers et al; Brothers).
Regarding claim 1 and analogous claims 13 and 14:
Brothers teaches:
1. A decoder for decoding weight parameters of a neural network, the decoder comprising:
(Brothers, ¶0048)
“FIG. 10 illustrates an embodiment in which the IDP decompressors for Huffman or Golomb-Rice decoding include a compressed weight mask stream decoder and a compressed weight value stream decoder [i.e. A decoder for decoding weight parameters of a neural network, the decoder comprising].”
2. a communication interface; a memory storing instructions; and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising;
(Brothers, ¶0049)
“Example embodiments can be deployed as an electronic device including a processor [i.e. and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising; ] and memory storing instructions [i.e. a memory storing instructions;]. Furthermore, it will be appreciated that embodiments can be deployed as a standalone device or deployed by multiple devices in distributed client-server networked system [i.e. a communication interface;].”
3. acquiring a plurality of bits representing weight parameters of the neural network;
(Brothers, ¶0030)
“FIG. 6 illustrates a method including pruning and retraining in accordance with an embodiment. Features maps and weights of a trained neural network are received 601 [i.e. representing weight parameters of the neural network;].”
(Brothers, ¶0037)
“The deltas and the base prediction are then compressed 650. A coding scheme, such an entropy coding scheme, may be used. For example, Huffman coding may be used represent the deltas with a number of bits [i.e. acquiring a plurality bits].”
3. decoding the weight parameters of the neural network using a context-dependent arithmetic coding;
(Brothers, ¶0048)
“FIG. 10 illustrates an embodiment in which the IDP decompressors for Huffman or Golomb-Rice decoding [i.e. decoding the weight parameters of the neural network using a context-dependent arithmetic coding;] include a compressed weight mask stream decoder and a compressed weight value stream decoder. In one embodiment, weight kernels are represented with masks specifying (pruned) weights and indices for non-zero weights."
Examiner notes that the specification does not define clearly what a “context-dependent arithmetic coding” is. In ¶0036 of the instant specification, an example is given wherein: “the context-dependent arithmetic coding can for example be a context-adaptive binary arithmetic coding, CABAC, wherein, also as an example, probabilities of bin values are determined for different contexts, and wherein, for example, each bin is associated with a context.”
4. selecting a context for a decoding of a weight parameter, or for a decoding of a syntax element of a number representation of the weight parameter, in dependence on one or more previously decoded weight parameters and/or in dependence on one or more previously decoded syntax elements of a number representation of one or more weight parameters;
(Brothers, ¶0047)
“After the reordering of the weight matrix, a set of Huffman tables is optimized for subsets of the nodes. For example, each table may correspond to a different set of nodes, with each table having a different frequency of low indices [i.e. selecting a context for a decoding of a weight parameter,].”
(Brothers, ¶0046)
“A single Huffman table is used to exploit the higher frequency of low indices throughout the whole set of weights. However, it is assumed in FIG. 9A that there is an even distribution of weight index usage—low indices are more common than high indices, but no more common in the left columns than the right ones [i.e. in dependence on one or more previously decoded weight parameters].”
5. decoding the weight parameter, or a syntax element of the weight parameter, using the selected context;
(Brothers, ¶0046)
“FIG. 10 illustrates an embodiment in which the IDP decompressors for Huffman or Golomb-Rice decoding include a compressed weight mask stream decoder and a compressed weight value stream decoder [i.e. decoding the weight parameter… using the selected context].”
6. and selecting a context out of two possible contexts for a decoding of a bit at a given bit position of the unary code or of the truncated unary code in dependence on a sign of the currently decoded weight parameter.
(Brothers, ¶0046)
“FIG. 10 illustrates an embodiment in which the IDP decompressors for Huffman or Golomb-Rice decoding include a compressed weight mask stream decoder and a compressed weight value stream decoder [i.e. and selecting a context out of two possible contexts for a decoding of a bit at a given bit position of the unary code].”
Regarding claim 15:
Brothers teaches:
1. An encoder for encoding weight parameters of a neural network, the encoder comprising:
(Brothers, ¶0045)
“As previously discussed, a number of different data compression algorithms can be used for the weights, such as, but not limited to, Huffman coding or any other suitable compression algorithm, such as Golomb-Rice coding [i.e. An encoder for encoding weight parameters of a neural network, the encoder comprising:].”
2. a communication interface; a memory storing instructions; and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising:
(Brothers, ¶0049)
“Example embodiments can be deployed as an electronic device including a processor [i.e. and a processor which is coupled to the communication interface and the memory, to execute the instructions stored in the memory to implement operations comprising; ] and memory storing instructions [i.e. a memory storing instructions;]. Furthermore, it will be appreciated that embodiments can be deployed as a standalone device or deployed by multiple devices in distributed client-server networked system [i.e. a communication interface;].”
3. obtaining a plurality of weight parameters of the neural network;
(Brothers, ¶0019)
“The feature maps and weights of the trained neural network are received 403 [i.e. obtaining a plurality of weight parameters of the neural network;].”
4. encoding the weight parameters of the neural network using a context-dependent arithmetic coding;
(Brothers, ¶0022)
“Still another option is to perform the reordering based on characteristics of a coding technique used for compression, such as Huffman coding or Golomb-Rice coding [i.e. encoding the weight parameters of the neural network using a context-dependent arithmetic coding;].”
5. selecting a context for an encoding of a weight parameter, or for an encoding of a syntax element of a number representation of the weight parameter, in dependence on one or more previously encoded weight parameters and/or in dependence on one or more previously encoded syntax elements of a number representation of one or more weight parameters;
(Brothers, ¶0047)
“After the reordering of the weight matrix, a set of Huffman tables is optimized for subsets of the nodes [i.e. in dependence on one or more previously encoded weight parameters]. For example, each table may correspond to a different set of nodes, with each table having a different frequency of low indices [i.e. selecting a context for an encoding of a weight parameter,]”
6. encoding the weight parameter, or a syntax element of the weight parameter, using the selected context;
(Brothers, ¶0022)
“Still another option is to perform the reordering based on characteristics of a coding technique used for compression, such as Huffman coding or Golomb-Rice coding.”
6. and selecting a context out of two possible contexts for a decoding of a bit at a given bit position of the unary code or of the truncated unary code in dependence on a sign of the currently encoded weight parameter.
(Brothers, ¶0046)
“FIG. 10 illustrates an embodiment in which the IDP decompressors for Huffman or Golomb-Rice decoding include a compressed weight mask stream decoder and a compressed weight value stream decoder [i.e. and selecting a context out of two possible contexts for a decoding of a bit at a given bit position of the unary code].”
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 2-12, 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over US Pre-Grant Patent 2018/0082181 (Brothers et al; Brothers) in view of “Golomb coding,” Wikipedia, further in view of “Entropy Coding in HEVC,” Springer-Verlag: MIT Libraries, Marpe et al; Marpe.
Regarding claim 2:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. selecting a context for the decoding of a zero flag of the weight parameter in dependence on a sign of a previously decoded weight parameter.
(Marpe, pg. 16, Sect. 3.2, ¶1, Fig. 3)
“Fig. 3: Context speculation required to achieve 5× parallelism when processing the significance map in H.264/AVC. Notation: i = coefficient position; i1 = MaxNumCoeff(BlockType)−1; EOB = end of block; SIG =significant coeff flag; LAST = last significant coeff flag [selecting a context for the decoding of a zero flag of the weight parameter in dependence on a sign of a previously decoded weight parameter].”
Examiner notes that the SIG flag is equivalent to a zero flag in the claim as it signals whether a coefficient is zero or non-zero. The context model selection therefore makes for the current zero/non-zero decision in dependence on information about the sign of previously decoded coefficients. It is further noted that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 3:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. determining a plurality of status identifiers represents statuses of a plurality of weight parameters at a plurality of positions relative to a position of a currently decoded weight parameter in the form of a numeric value,
(Marpe, pg. 35-pg.36, Sect. 6.4, ¶3)
“In H.264/AVC, the significance map for each transform block is signaled by transmitting a significant coeff flag (SIG) for each position to indicate whether the coefficient is non-zero [i.e. determining a plurality of status identifies represents statuses of a plurality of weight parameters]. The positions are processed in an order based on a zig-zag scan [i.e. at a plurality of positions relative to a position of a currently decoded weight parameter]. After each non-zero SIG, an additional flag called last significant coeff flag (LAST) is immediately sent to indicate whether it is the last non-zero SIG; this prevents unnecessary SIG from being signaled [i.e. in the form of a numeric value,].
2. and combining the status identifiers, in order to acquire a context index value representing a context of the currently decoded weight parameter.
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. in order to acquire a context index value representing a context of the currently decoded weight parameter].”
PNG
media_image1.png
167
540
media_image1.png
Greyscale
Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 4:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. selecting a context for the decoding of a zero flag of the weight parameter in dependence on how many zero-valued weight parameters and/or unavailable weight parameters in a row are adjacent to the currently decoded weight parameter.
(Marpe, pg. 39, Sect. 6.4.2., ¶1))
“If all of the N SIG are zero, LAST is not transmitted. [73] avoids interleaving of SIG and LAST altogether. Specifically, the horizontal (x) and vertical (y) position of the last non-zero SIG in a TB is sent in advance rather than LAST by using the syntax elements last sig coeff x and last sig coeff y, respectively [i.e. selecting a context for the decoding of a zero flag of the weight parameter in dependence on how many zero-valued weight parameters].”
Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 5:
Brothers teaches the machine of claim 1.
Brothers teaches:
1. wherein the plurality of weight parameters is arranged in a matrix, and the weight parameters are denoted as (Ix-1,y), (Ix-2,y), (Ix-3,y), and correspond to positions (x-1,y), (x- 2,y) and (x-3,y) in the matrix, respectively,
(Brothers, ¶0047, Fig. 9B)
“In the example of FIG. 9B, the reordering moves low-value weights to one side of a matrix and high values to the other side [i.e. wherein the plurality of weight parameters is arranged in a matrix, and the weight parameters are denoted as (Ix-1,y), (Ix-2,y), (Ix-3,y)]. After the reordering of the weight matrix [i.e. correspond to positions (x-1,y), (x- 2,y) and (x-3,y) in the matrix, respectively,], a set of Huffman tables is optimized for subsets of the nodes.”
Marpe teaches:
1. and are represented by status identifiers Sx-1 ,y, Sx-2,y, Sx-3,y.
(Marpe, pg. 35, Sect. 6.4, ¶3)
“In H.264/AVC, the significance map for each transform block is signaled by transmitting a significant coeff flag (SIG) for each position to indicate whether the coefficient is non-zero [i.e. and are represented by status identifiers Sx-1 ,y, Sx-2,y, Sx-3,y].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 6:
Brothers teaches the machine of claim 1.
Brothers teaches:
1. wherein the plurality of weight parameters is arranged in a matrix,
(Brothers, ¶0047, Fig. 9B)
“In the example of FIG. 9B, the reordering moves low-value weights to one side of a matrix and high values to the other side [i.e. wherein the plurality of weight parameters is arranged in a matrix,].”
Marpe teaches:
1. and a status identifier s,y for a position (x,y) in the matrix is equal to a first value, if the position (x,y) is not available or the weight parameter at the position (x,y) is equal to zero,
PNG
media_image2.png
429
524
media_image2.png
Greyscale
(Marpe, pg. 31, Sect. 6, ¶1)
“As a reference for the beginning and end points of the development, Fig. 8 and Fig. 9 show examples of transform coefficient coding for 4×4 blocks in H.264/AVC and HEVC, respectively [i.e. or the weight parameter at the position (x,y) is equal to zero,].”
PNG
media_image1.png
167
540
media_image1.png
Greyscale
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. and a status identifier s,y for a position (x,y) in the matrix is equal to a first value].”
Examiner notes that Fig. 9 shows that the transform coefficients, analogous to the weight coefficients in a neural network, can be zero. The subsequent status identifier for the Fig. 12 is equal to a first value.
PNG
media_image2.png
429
524
media_image2.png
Greyscale
2. the status identifier for the position (x,y) is equal a second value, if the weight parameter at the position (x,y) is smaller than zero,
(Marpe, pg. 31, Sect. 6, ¶1)
“As a reference for the beginning and end points of the development, Fig. 8 and Fig. 9 show examples of transform coefficient coding for 4×4 blocks in H.264/AVC and HEVC, respectively [i.e. if the weight parameter at the position (x,y) is smaller than zero,].”
PNG
media_image1.png
167
540
media_image1.png
Greyscale
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. the status identifier for the position (x,y) is equal a second value,].”
Examiner notes that Fig. 9 shows that the transform coefficients, analogous to the weight coefficients in a neural network, can be non-zero and negative. The subsequent status identifier for the Fig. 12 is equal to a second value.
PNG
media_image2.png
429
524
media_image2.png
Greyscale
(Marpe, pg. 31, Sect. 6, ¶1)
“As a reference for the beginning and end points of the development, Fig. 8 and Fig. 9 show examples of transform coefficient coding for 4×4 blocks in H.264/AVC and HEVC, respectively [i.e. if the weight parameter at the position (x,y) is larger than zero,].”
PNG
media_image1.png
167
540
media_image1.png
Greyscale
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. and the status identifier for the position (x,y) is equal to a third value,”
Examiner notes that Fig. 9 shows that the transform coefficients, analogous to the weight coefficients in a neural network, can be non-zero and positive. The subsequent status identifier for the Fig. 12 is equal to a third value.
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 7:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. wherein the plurality of weight parameters is arranged in a matrix, and a status identifier s,y for a position (x,y) in the matrix is equal to a first value, if the position (x,y) is not available or the weight parameter at the position (x,y) is equal to zero,
PNG
media_image1.png
167
540
media_image1.png
Greyscale
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. wherein the plurality of weight parameters is arranged in a matrix, and a status identifier s,y for a position (x,y) in the matrix is equal to a first value,…. or the weight parameter at the position (x,y) is equal to zero,]”
2. and the status identifier sX,y for the position (x,y) is equal to a second value, if the position (x,y) is available and the weight parameter at the position (x,y) is not equal to zero.
PNG
media_image1.png
167
540
media_image1.png
Greyscale
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. and the status identifier sX,y for the position (x,y) is equal to a second value, if the position (x,y) is available and the weight parameter at the position (x,y) is not equal to zero].”
Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 8:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. selecting a context for the decoding of a zero flag of the weight parameter in dependence on a distance of a closest non-zero weight parameter present in a predetermined direction, when seen from the currently decoded weight parameter.
(Marpe, pg. 36, Sect. 6.4.1., ¶2, Fig. 13a-13f)
“Specifically, the context selection of SIG was calculated based on a local template using 10 (already decoded) SIG neighbors as shown in Fig. 13a [102, 59]. By using this template-based context selection bitrate savings of 1.4 –2.8% were reported [59].”
PNG
media_image3.png
580
597
media_image3.png
Greyscale
Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 9:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. selecting a context for the decoding of a zero flag of the weight parameter considering only a single one previously decoded weight parameter, which is adjacent to the currently decoded weight parameter.
(Marpe, pg. 36, Sect. 6.4.1, ¶4)
“Despite reducing the number of SIG neighbors in HM2.0, dependency on the most recently processed SIG neighbors still existed for the positions at the edge of the transform block as shown in Fig. 14a [i.e. selecting a context for the decoding of a zero flag of the weight parameter considering only a single one previously decoded weight parameter,]. The horizontal or vertical shift that is required to go from one diagonal to the next in the zig-zag scan causes the previously decoded bin to be one of the neighbors (F or I) that is needed for context selection [i.e. which is adjacent to the currently decoded weight parameter].”
Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 10:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. determining a status identifier for the single one previously decoded weight position, wherein the status identifier for the single one previously decoded weight parameter equals to a first value,
(Marpe, pg. 35-pg.36, Sect. 6.4, ¶3)
“In H.264/AVC, the significance map for each transform block is signaled by transmitting a significant coeff flag (SIG) for each position to indicate whether the coefficient is non-zero [i.e. determining a status identifier for the single one previously decoded weight position, wherein the status identifier for the single one previously decoded weight parameter equals to a first value,]
2. if the single one previously decoded weight parameter is not available or the weight parameter at the position (x,y) is equal to zero, equals to a second value
(Marpe, pg. 39, Sect. 6.4.2., ¶1)
“If all of the N SIG are zero, LAST is not transmitted [i.e. the weight parameter at the position (x,y) is equal to zero, equals to a second value].”
3. if the single one previously decoded weight parameter is smaller than zero,
PNG
media_image2.png
429
524
media_image2.png
Greyscale
(Marpe, pg. 31, Sect. 6, ¶1, Fig. 9)
“As a reference for the beginning and end points of the development, Fig. 8 and Fig. 9 show examples of transform coefficient coding for 4×4 blocks in H.264/AVC and HEVC, respectively [i.e. if the single one previously decoded weight parameter is smaller than zero,].”
4. and equals to a third values, if the single one previously decoded weight parameter is larger than 0;
PNG
media_image2.png
429
524
media_image2.png
Greyscale
(Marpe, pg. 31, Sect. 6, ¶1, Fig. 9)
“As a reference for the beginning and end points of the development, Fig. 8 and Fig. 9 show examples of transform coefficient coding for 4×4 blocks in H.264/AVC and HEVC, respectively [i.e. and equals to a third values, if the single one previously decoded weight parameter is larger than 0;].”
5. and selecting the context in dependence on the status identifier.
(Marpe, pg. 36, Sect. 6.4.1., ¶2)
“While in HEVC position based context assignment for coding of sig coeff flag (SIG) is used for 4×4 TBs as shown in Fig. 12, new forms of context assignment for larger transforms were needed [i.e. and selecting the context in dependence on the status identifier].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 11:
Brothers teaches the machine of claim 1.
Marpe teaches:
1. selecting a context associated with a zero value of the previously decoded weight parameter in case the previously decoded weight parameter is not available.
(Marpe, pg. 29, Sect. 5.2, ¶3)
“In HEVC, additional MPMs are used to improve coding efficiency. A candidate list of most probable modes with a fixed length of three is constructed based on the left and top neighbors [i.e. the previously decoded weight parameters]. The additional candidate modes (DC, planar, vertical) can be added if the left and top neighbors are the same or unavailable [i.e. selecting a context associated with a zero value of the previously decoded]. Note that the top neighbors outside current CTU are considered unavailable in order to avoid the need for a line buffer [i.e. in case… is not available].”
Examiner notes that in the claim’s terms, the “previously decoded” corresponds to a neighboring mode in the MPM process. The term “associated with a zero value” aligns with the default or fallback assumption when no prior value is present. Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 12:
Brothers teaches the machine of claim 1.
Brothers teaches:
1. wherein the weight parameters are organized in rows and columns of a matrix, wherein an order in which the weight parameters are decoded is along a first row of the matrix, then along a subsequent second row of the matrix, or wherein an order in which the weight parameters are decoded is along a first column of the matrix, then along a subsequent second column of the matrix.
(Brothers, ¶0035)
“The deltas are computed versus prediction 635. For example, the differences between adjacent columns and/or rows in a cluster may be computed [i.e. wherein the weight parameters are organized in rows and columns of a matrix, wherein an order in which the weight parameters are decoded is along a first row of the matrix, then along a subsequent second row of the matrix,].”
Examiner further notes that the term “weight parameters” is taught by the Brothers reference, “FIG. 1 is a high level block diagram in accordance with an embodiment. In one embodiment, a neural network (NN) development framework 105 generates a set of weights for all of the layers of the network [Brothers, 0014].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 16:
Brothers teaches the machine of claim 1.
Brothers teaches:
1. A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using a decoder according to claim 1
(Brothers, ¶0018)
“FIG. 3 shows in more detail an example of some of the data streams that feed the MAA units in accordance with an embodiment… In one embodiment, an individual IDP provides one non-zero weight to a MAA and one IFM (e.g., a 4×4 block) to each of the MAAs [i.e. A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using a decoder according to claim 1].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Regarding claim 17:
Brothers teaches the machine of claim 1.
Brothers teaches:
1. A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using an encoder according to claim 15.
(Brothers, ¶0018)
“FIG. 3 shows in more detail an example of some of the data streams that feed the MAA units in accordance with an embodiment… In one embodiment, an individual IDP provides one non-zero weight to a MAA and one IFM (e.g., a 4×4 block) to each of the MAAs [i.e. A non-transitory computer-readable medium for storing data associated with a data stream, wherein the data stream comprising weight parameters of the neural network is encoded thereinto using an encoder according to claim 15.].”
One of ordinary skill in the art, at the time the invention was filed, would have been motivated to modify Brothers with Marpe. The motivation is the same as claim 1.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 PAUL JUSTIN BREENE whose telephone number is (571)272-6320. 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, Michael J Huntley can be reached on 303-297-4307. 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.
/P.J.B./ Examiner, Art Unit 2129
/MICHAEL J HUNTLEY/Supervisory Patent Examiner, Art Unit 2129
Prosecution Insights