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 .
The office action is in response to the application filed on October 11, 2023.
Claims 1-20 are pending and have been examined. Claims 1-20 are rejected.
Information Disclosure Statement
Acknowledgment is made of the information disclosure statements filed December 14, 2023, October 29, 2026, and September 16, 2023 which comply with 37 CFR 1.97. As such, the information disclosure statements have been placed in the application file and the information referred to therein has been considered by the examiner.
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.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1, 8, and 15 are rejected under 35 U.S.C. 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding Claims 1, 8, and 15, the limitations "the SNN comprises accuracy-latency balance (ALB) characteristics" AND "achieves a predetermined ALB of an output generated by the SNN" do not clearly set the metes and bounds of the patent protection desired. The term "ALB characteristics" is a relative and result-functional term which renders the claim indefinite, as it defines the claimed SNN entirely by the outcome produced rather than by any structural, operational, or algorithmic feature that constitutes those characteristics. Likewise, the term “predetermined ALB” is indefinite, as neither the claim nor the specification establishes a fixed metric, unit, or comparison standard by which “ALB” is measured or by which “achievement” of a “predetermined” value could be evaluated. The specification does not provide a standard for ascertaining the requisite degree of accuracy-latency balance, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention in regards to what structural characteristics are required of the SNN, nor what quantity or threshold must be met to satisfy “predetermined” further rendering the claims indefinite. To expedite prosecution under BRI, the examiner will interpret and liken “ALB characteristics” to any feature, parameter, or configuration of the SNN including but not limited to weight values, threshold values, or training/mapping procedures that has any bearing on the accuracy and/or latency of the SNN’s output, and/or where it applies. Likewise, “predetermined ALB” will be interpreted as any accuracy and/or latency value or range selected, targeted, or configured prior to execution of the SNN, and/or where it applies.
Claim Rejections - 35 USC § 102
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 (i.e., changing from AIA to pre-AIA ) 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 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)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1, 8, and 15 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by
Kim et. Al, (Towards Fast and Accurate Object Detection in Bio-Inspired Spiking Neural Networks Through Bayesian Optimization, published December 24, 2020 hereinafter "Kim.S"). whose date is before the effective filing date of this application, i.e., December 14, 2023. Therefore, Kim.S constitutes prior art under 35 U.S.C. 102(a)(1).
With respect to independent Claims 1, 8, and 15:
Kim.S teaches:
“executing a spiking neural network (SNN) to perform an SNN task;” ([pg. 2637 Step. 1] discloses executing a spiking neural network to perform object recognition/detection tasks, “SNNs are now executed for T time steps and produce object detection accuracy (mAP).” [pg. 2640 sec. 4] further discloses the specific configuration of the spiking neural network that will be necessary for fulfilling the SNN task (object detection), “Spike-based object detection model used in this work is based on Spiking-YOLO [22] which utilizes a real-time object detection model called Tiny YOLO.”)
“wherein the SNN comprises accuracy-latency balance (ALB) characteristics that enable the SNN to perform the SNN task in a manner---" ([pg. 2639, sec. B.1] teaches accuracy-latency balancing characteristics/built-in SNN configuration (two-phase threshold voltage scheme) that allows the SNN to perform the SNN task (object detection), “The threshold voltages in each phase aim to achieve two distinct goals: phase-1 threshold voltages for fast object detection and phase-2 threshold voltages for accurate object detection. Phase-1 threshold voltages primarily focus on transmitting information fast and early, reducing latency. This may, however, result in rough estimate detection. Phase-2 threshold voltages, on the other hand, concentrate on achieving accurate object detection. This may result in high latency. Combining the two, we can achieve faster and more accurate object detection in SNNs as opposed to conventional methods.” [pg. 2638, sec. 3] further discloses the implementation of these ALB characteristics, “a spiking neuron whose threshold voltage is 0.7 V, will be more likely to generate a spike since a lower amount of integrated membrane potential is required to reach the threshold voltage of 0.7 V. Contrarily, spiking neurons with a threshold voltage greater than the baseline will less likely generate a spike since more membrane potential needs to be integrated before generating a spike.”)
“---that achieves a predetermined ALB of an output generated by the SNN." ([pg. 2640, sec. 5] discloses an instance of predetermined ALB criteria fixed in advance and established before outputs are generated by the SNN, “The target mAP of Tiny YOLO is 53.01% (PASCALVOC) and 26.24% (MS COCO). As demonstrated in Figure 7 (a), in PASCAL VOC, the proposed method achieved an object detection accuracy of 51.45% at 5,000-time steps. More importantly, to reach 95% of the DNN’s target accuracy (53.01%), only 1,300-time steps are required compared with 3,400-time steps at the baseline (2.6x faster).”) Examiner’s Note: Fixed/predetermined criteria precede Bayesian optimization search for threshold-voltage configurations and is thus, determining/measuring accuracy and latency based on said target.
Therefore, Claims 1, 8, and 15 are rejected.
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 (i.e., changing from AIA to pre-AIA ) 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.
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.
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 non-obviousness.
Claims 2, 3, 9, 10, 16, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Kim et. Al, (Towards Fast and Accurate Object Detection in Bio-Inspired Spiking Neural Networks Through Bayesian Optimization, published December 24, 2020, hereinafter "Kim.S"), in view of Park et. Al, (T2FSNN: Deep Spiking Neural Networks with Time-to-first-spike Coding, published March 26, 2020, hereinafter "Park"). The dates are before the effective filing date of this application, i.e., December 14, 2023 where it applies.
With respect to Claims 2, 9, and 16:
Park teaches:
"wherein the ALB characteristics are based at least in part on an accuracy and latency function (ALF).” ([pg. 3, sec. B] discloses the common relationship between the ALB characteristics and functions that determine both accuracy and latency, “Thus, there is a trade-off between the precision and latency of information transmission between neurons, depending on the time constant τ in the kernel. To address such a trade-off properly, we propose loss functions, including precision loss and representation loss, considering the accuracy and latency of the inference. In addition, we propose a gradient-based optimization, which is based on supervised learning and minimizes the loss functions in a layer-wise manner.”)
Kim.S-Park are analogous art and in the same field of invention because both references pertain to optimizing spiking neural network (SNN) efficiency to reduce inference latency, computational overhead, and synaptic operations for practical real-world deployment. While Kim.S teaches dynamically balancing threshold voltages specifically for object detection, Park teaches utilizing time-to-first-spike (T2FS) coding with a kernel-based threshold to convert deep networks for broader practical tasks. It would have been obvious to a person having ordinary skill in the art (PHOSITA) before the effective filing date of the claimed invention to implement the base reference of Kim.S (calculating precise bounding boxes using Bayesian optimization) with the teachings of Park (improving SNN applicability by handling spatial and temporal data) in order to build an ultrafast, highly accurate, and extremely low-power AI model that can run efficiently on neuromorphic hardware. One of ordinary skill in the art would be motivated to do so because by integrating Park's framework into the methods of Kim.S one would be able to note that a system implementation as such, "achieved 46.9% reduction in inference latency while improving accuracy of 0.07% and 2.75% on CIFAR-10 and CIFAR-100, respectively. Furthermore, the number of spikes also decreased by 0.3% and 2.1%. It is interesting to note that the accuracy was improved with a lower number of spikes, despite that early firing causes non-guaranteed integration, {[pg. 5 para. 2] of Park}."
Therefore, Claims 2, 9, and 16 are rejected.
With respect to Claims 3, 10, and 17:
Park teaches:
"wherein the ALB characteristics are based at least in part on a set of ALB terms applied to the ALF.” ([pg. 4, Fig. 3] discloses a set of accuracy-latency balancing terms applied to the accuracy-latency loss functions, “Pipeline of the integration and fire phase The representation loss consists of two terms; Ll min and Ll max. These two loss terms consider the minimum and maxi mum representation values of each layer’s kernel, so that the kernel can learn the distribution of ground truth.” [pg. 4, sec. A] further details execution of the ALB terms with respect to the ALF, “To evaluate the proposed optimization method, we measured the three loss terms (Lprec, Lmin, and Lmax). We set two different initial conditions to validate the trade-off between precision and latency of information transmission depending on the τ: 1) a small time constant (τ=2), and 2) a large time constant (τ=18) on a given time window (T=20). When the time constant is a small value (τ=2), the kernel can represent small values sufficiently, but the precision of transmission is low. Thus, as the training progresses, the time constant τ increased and the precision loss Lprec decreased as shown in Fig. 4-(a).”) Examiner’s Note: ALB terms utilized with respect to gradient-based optimization methods to obtain minimum representation loss [equation 13].
Therefore, Claims 3, 10, and 17 are rejected.
Claims 4, 11, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Kim.S, in view of Park, in further view of Kheradpisheh et. Al, (S4NN: temporal backpropagation for spiking neural networks with one spike per neuron, published June 13, 2020, hereinafter “Kheradpisheh”), in further view of Kim et. Al, (Rethinking Skip Connections in Spiking Neural Networks with Time-To-First-Spike Coding, published December 1, 2023, hereinafter “Kim.Y”). The dates are before the effective filing date of this application, i.e., December 14, 2023 where it applies.
With respect to Claims 4, 11, and 18:
The combination of Kim.S-Park alone does not appear to explicitly disclose:
"wherein: the set of ALB terms comprises a regularization loss term and a cross-entropy loss term; and the ALF comprises a set of model training operations.”
However, Kheradpisheh teaches:
PNG
media_image1.png
39
143
media_image1.png
Greyscale
"wherein: the set of ALB terms comprises a regularization loss term and-----” ([pg.5, sec. 2.4] discloses utilizing a regularization loss term, “To avoid over-fitting, we added an L2-norm regularization term (over all the synaptic weights in all the layers) to the “squared error” loss function in Eq. (10). The parameter λ is the regularization parameter accounting for the degree of weigh penalization.”)
Kim.S-Park-Kheradpisheh are analogous art and in the same field of invention because all three references pertain to creating highly efficient, fast, and scalable bio-inspired neural models that can transmit information using sparse, discrete events (spikes) rather than continuous values. While Kim.S teaches mimicking the human brain to achieve sparse, energy-efficient information transmission using spike trains, Park teaches bypassing the inefficiencies of standard network-to-SNN conversion techniques with the use of a dendrite model. Similarly, Kheradpisheh implements rank-order coding where exactly one spike is fired per neuron with the first neuron to fire in a readout layer determining the classification. It would have been obvious to a person having ordinary skill in the art (PHOSITA) before the effective filing date of the claimed invention to implement the base reference of Kim.S (fine-tuning network parameters with a Gaussian process) with the teachings of Park (replacing rate coding with sparse temporal coding by relying only on the first spike a neuron emits) and the teachings of Kheradpisheh (backpropagation adapted for spike latencies) in order to prevent the historical difficulties of training SNNs directly by redefining how they encode information and process events (e.g., using latencies or firing orders). One of ordinary skill in the art would be motivated to do so because by integrating Park and Kheradpisheh's frameworks into the methods of Kim.S one would be able to recognize that, "by increasing the threshold, the accuracy increases, goes above 94% after threshold 70, and peaks at the threshold 100. Also, it can be seen that the mean response-time fastly grows after threshold 70, {[pg. 12] of Kheradpisheh}."
Kheradpisheh alone does not appear to explicitly disclose:
“---a cross-entropy loss term; and the ALF comprises a set of model training operations.”
However, Kim.Y teaches:
PNG
media_image2.png
67
293
media_image2.png
Greyscale
“---a cross-entropy loss term; and the ALF comprises a set of model training operations.” ([pg. 9, sec. 3.4] discloses a cross-entropy loss term used to calculate accuracy and latency, “In the output layer, the class probability is computed from the spike timing. The objective is to train the network such that the neuron associated with the correct class is the first to fire among all neurons. This can be done by applying the cross-entropy loss to the output spike timing O ∈ RC of the last layer: …With a cross-entropy loss for classification, following the prior research [17], we introduce an additional term to the cost function that penalizes the input weight vectors of neurons whose sum is below 1 (denominator of Eq. 6).” [pg. 10, sec. 4.2] further discloses model training operations, “To compare these architectures, we introduce latency as an additional metric, along with accuracy. In this context, latency is defined as the average time taken for the first spike to occur in the final layer. This is a particularly relevant measure for TTFS coding, as operations can be terminated as soon as the first spike occurs in the last layer. As such, we present both latency and accuracy in our results, offering a comprehensive understanding of the trade-off between speed and precision in these varying architectural designs. In Table 1 and Table 4.2, we present the accuracy and latency results for the MNIST and Fashion-MNIST datasets respectively.” [pg. 17, sec. Dataset Configuration] further discloses the ALF (wave equation function) that encompasses multiple model training operations as well as measures accuracy and latency, “Finally, we partition these data samples randomly into training and testing sets at an 80:20 ratio, respectively. In Table 4.5, we report the accuracy and latency of the wave equation problem.”) Examiner’s Note: [Table 7] reports the accuracy-latency pairing for the wave equation (ALF) and the results of the model training operations.
Kim.S-Park-Kheradpisheh-Kim.Y are analogous art and in the same field of invention because all four references pertain to overcoming the computational bottlenecks of traditional SNNs specifically high energy costs, latency, and training limits by leveraging precise temporal coding and improved network architectures. While Kim.S teaches dynamically balancing network thresholds with the benefit of improving real-time object detection precision, Park teaches utilizing spatio-temporal information to develop highly efficient deep SNNs that process data faster. Similarly, Kheradpisheh teaches minimizing spike count by implementing supervised learning in multilayer SNNs akin to traditional backpropagation, while Kim.Y teaches integrating standard deep-learning architectural concepts with Time-To-First-Spike (TTFS) coding to improve information flow and energy efficiency. It would have been obvious to a person having ordinary skill in the art (PHOSITA) before the effective filing date of the claimed invention to implement the base reference of Kim.S (threshold balancing) with the teachings of Park (T2FSNN) with the teachings of Kheradpisheh (S4NN) further with the teachings of Kim.Y (skip connections) in order to overcome the fundamental trade-off between high performance/accuracy and low energy/latency in SNNs. One of ordinary skill in the art would be motivated to do so because by integrating Park, Kheradpisheh, and Kim.Y’s frameworks into the methods of Kim.S one would be able to recognize that a system as such, "maintains accuracy, offering an effective solution for faster prediction in TTFS coding, {[pg. 18] of Kim.Y}."
Therefore, Claims 4, 11, and 18 are rejected.
Claims 5, 6, 12, 13, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Kim.S, in view of Park, in further view of Rueckauer et. Al, (Conversion of Continuous-Valued Deep Networks to Efficient Event-Driven Networks for Image Classification, published December 06, 2017, hereinafter “Rueckauer”). The dates are before the effective filing date of this application, i.e., December 14, 2023 where it applies.
With respect to Claims 5, and 12:
The combination of Kim.S-Park alone does not appear to explicitly disclose:
"wherein: the set of ALB terms comprises ALB mapping terms; and the ALB mapping terms comprise a hyperparameter.”
However, Rueckauer teaches:
"wherein: the set of ALB terms comprises ALB mapping terms; and the ALB mapping terms comprise a hyperparameter.” ([pg. 5, sec. 2.2.2.2] discloses that the accuracy-latency balancing terms also consist of mapping terms that are hyperparameters, “We propose a more robust alternative where we set λl to the p-th percentile of the total activity distribution of layer l4. This choice discards extreme outliers, and increases SNN firing rates for a larger fraction of samples... so choosing the normalization scale involves a trade-off between saturation and insufficient firing. In the following, we refer to the percentile p as the “normalization scale,” and note that the “max-norm” method is recovered as the special case p = 100. Typical values for p that perform well are in the range [99.0, 99.999].”) Examiner’s Note: The percentile p, defining the λl normalization scale factor used during the ANN-to-SNN conversion, functions as a hyperparameter because its selection involves a manual trade-off between neuron saturation and insufficient firing. Essentially, rescales the ANN’s weights and biases so the SNN firing threshold aligns with the ANN’s activation distribution (mapping).
Kim.S-Park-Rueckauer are analogous art and in the same field of invention because all three references pertain to achieving high computational efficiency and lower energy consumption without sacrificing the accuracy of spiking neural networks. While Kim.S teaches achieving higher detection accuracy without sacrificing the sparse energy-saving nature of SNNs, Park teaches finding a scalable solution to standard SNN conversion issues. Similarly, Rueckauer develops spiking equivalents for complex operations like batch-normalization and Inception-modules. It would have been obvious to a person having ordinary skill in the art (PHOSITA) before the effective filing date of the claimed invention to implement the base reference of Kim.S (voltage adjustments) with the teachings of Park (lowering inference latency) and the teachings of Rueckauer (deep SNN architecture conversion) in order to close the gap between traditional continuous-valued deep networks and SNNs by improving inference latency, energy use, and architectural flexibility. One of ordinary skill in the art would be motivated to do so because by integrating Park and Rueckauer's frameworks into the methods of Kim.S one would be able to recognize that, "these results were confirmed also on MNIST, where a 7-layer network with max-pooling achieved an error rate of 0.56%, thereby improving previous state-of-the-art results for SNNs reported by Diehl et al. (2015) and Zambrano and Bohte (2016), {[pg. 7] of Rueckauer}."
Therefore, Claims 5, and 12 are rejected.
With respect to Claims 6, and 13:
The combination of Kim.S-Park-Rueckauer teaches:
"wherein the ALF comprises a set of model mapping operations applied to an artificial neural network (ANN) model to create the SNN.” ([pg. 4, sec. 2.2.2.1] discloses a set of model mapping operations (weight/bias normalization) to facilitate the creation of an SNN, “The data-based weight normalization mechanism is based on the linearity of the ReLU unit used for ANNs. It can simply be extended to biases by linearly rescaling all weights and biases such that the ANN activation a [as computed in Equation (1)] is smaller than 1 for all training examples. In order to preserve the information encoded within a layer, the parameters of a layer need to be scaled jointly. Denoting the maximum ReLU activation in layer l as λl = max[al], then weights Wl and biases bl are normalized to Wl → Wlλl−1 λl andbl → bl/λl.” [pg. 2, col. 2] further discloses the creation of an SNN from the basis of an ANN, “To automate the process of transforming a pre-trained ANN into an SNN, we developed an SNN-conversion toolbox that is able to transform models written in Keras (Chollet, 2015), Lasagne and Caffe, and offers built-in simulation tools for evaluation of the spiking model.”)
Therefore, Claims 6, and 13 are rejected.
With respect to Claim 19:
Examiner’s Note: Repeated limitations. Refer to rejection rationale for claims 5 and 6.
Therefore, Claim 19 is rejected.
Claims 7, 14, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Kim.S, in view of Kheradpisheh. The dates are before the effective filing date of this application, i.e., December 14, 2023 where it applies.
With respect to Claims 7, 14, and 20:
Kim.S does not appear to explicitly disclose:
"wherein: the SNN comprises a single-spike SNN (SS-SNN).”
However, Kheradpisheh teaches:
"wherein: the SNN comprises a single-spike SNN (SS-SNN).” ([pg. 2, sec. 2] discloses that the SNN comprises a single-spike SNN, “The proposed single-spike supervised spiking neural network (S4NN) is comprised of an input layer converting input data into a spike train and feeding it into the network, followed by one or more hidden layers of non-leaky integrate-and-fire (IF) neurons processing the input spikes, and finally, an output layer of non-leaky IF neurons with one neuron per category.”)
Kim.S-Kheradpisheh are analogous art and in the same field of invention because both references pertain to developing computationally efficient energy-saving AI models for computer vision and edge computing by optimizing how networks process information through discrete time-based signals. While Kim.S teaches identifying the primary theoretical advantages of SNNs (exceptional computational and energy efficiency due to sparse spike-based transmission), Kheradpisheh proposes a supervised learning rule using rank-order coding and temporal backpropagation to mitigate systemic bottlenecks such as latency and synaptic operations. It would have been obvious to a person having ordinary skill in the art (PHOSITA) before the effective filing date of the claimed invention to implement the base reference of Kim.S (fine-tuning network parameters with a Gaussian process) with the teachings of Kheradpisheh (backpropagation adapted for spike latencies) in order to solve the inherent trade-off between model efficiency and learning capability/training speed. One of ordinary skill in the art would be motivated to do so because by integrating Kheradpisheh's frameworks into the methods of Kim.S one would be able to recognize that, "by increasing the threshold, the accuracy increases, goes above 94% after threshold 70, and peaks at the threshold 100. Also, it can be seen that the mean response-time fastly grows after threshold 70, {[pg. 12] of Kheradpisheh}."
Therefore, Claims 7, 14, and 20 are rejected.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure: Xiao et. Al (Optimal Mapping of Spiking Neural Network to Neuromorphic Hardware for Edge-AI) and Van Der Made et. Al (US20200143229AI).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NOOR F CHEEMA whose telephone number is (571)272-9642. The examiner can normally be reached Monday-Friday 7:30am-5:00pm alternative Fridays off.
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, Mariela Reyes can be reached at (571) 270-1006. 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.
/N.F.C./Examiner, Art Unit 2142
/Mariela Reyes/Supervisory Patent Examiner, Art Unit 2142
Prosecution Insights