Abstract: Transformer neural network based decoder for decoding Quantum Error Correction Codes (QECC), comprising, an input layer, a plurality of decoding layers, and an output layer. The input layer is adapted to receive initial noise estimation computed by a noise estimator for noise injected to syndrome bits of codewords encoded using QECC and transmitted over transmission channel(s) subject to interference, and create embeddings for the syndrome bits. The decoding layers adapted to compute an estimated logical operator matrix of each codeword, each comprises a self-attention layer constructed according to a mask indicative of a relation between the embeddings derived from a parity-check matrix of the error correction code. The plurality of decoding layers are trained using a combined loss function directed to minimize LER, BER, and error rate of the noise estimator. The output layer is adapted to produce a vector representing predicted soft error of the codeword's logical operator matrix.

Abstract: A computing device performs generating a first identity encoding representing a first facial identity of the person based on an image of a person, generating a second identity encoding representing a second facial identity different from the first facial identity of the person based on the first identity encoding, generating a source encoding by using an encoder to process a source image of the person having an expression, generating an intermediate image by using a decoder to process the source encoding and the second identity encoding, the intermediate image including a face having the second facial identity and the expression of the person in the source image, and generating an output image by blending the source image with facial features of the face in the intermediate image.

Abstract: The invention presents a universal foundation model for decoding error correction codes (ECC) and a learning-based method for linear block ECCs. The foundation model is trained on multiple codes using a code-invariant embedding, relative positional encoding from derived from parity-check matrices, and a size-invariant transformation to generate a robust noise prediction. A learned distance embedding derived from each code's Tanner graph modulates self-attention, allowing the system to handle both seen and unseen codes without retraining. Overall, this approach replaces specialized, code-specific decoders with a single efficient model, enabling more robust and scalable decoding of diverse error correction codes. The linear block learning component based on the Transformer architecture allows the differentiable training of the code via the Tanner graph connectivity derivation from the parity check matrix, and enables the effective and differentiable joint optimization of the code and of the neural decoder.

Abstract: Disclosed herein are systems and method for training neural network based decoders for decoding error correction codes, comprising obtaining a plurality of training samples comprising one or more codewords encoded using an error correction code and transmitted over a transmission channel where the training samples are subject to gradual interference over a plurality of time steps and associate the encoded codeword(s) with an interference level and a parity check syndrome at each of the plurality of time steps, using the training samples to train a neural network based decoder to decode codewords encoded using an error correction code by (1) estimating a multiplicative interference included in the encoded codeword(s) based on reverse diffusion applied to the encoded codeword(s) across the time steps, (2) computing an additive interference included in the encoded codewords based on the multiplicative interference, and (3) recovering the codeword(s) by removing the additive interference.

Abstract: A method of designing a nanostructure, comprises: receiving a far field optical response and material properties; feeding the synthetic far field optical response and material properties to an artificial neural network having at least three hidden layers; and extracting from the artificial neural network a shape of a nanostructure corresponding to the far field optical response.

Abstract: The disclosure comprises a method to improve machine learning models by cleaning training data using anomaly detection, as well as anomaly detection per se. The method considers the task of finding out-of-class samples in tabular data, where little may be safely assumed about the structure of the data. The method captures the structure of the samples of the single training class, by learning mappings that maximize the mutual information between each sample and a part that is masked out. The mappings are learned by employing a contrastive loss that considers only one sample at a time. Once learned, the disclosure may score a test sample by measuring whether the learned mappings lead to a small contrastive loss using the masked parts of this sample. The experiments show accuracy advantage in comparison to the literature using the same set of hyperparameters as the state of the art results across benchmarks.

Abstract: Disclosed herein are systems and method for training neural network based decoders for decoding error correction codes, comprising obtaining a plurality of training samples comprising one or more codewords encoded using an error correction code and transmitted over a transmission channel where the training samples are subject to gradual interference over a plurality of time steps and associate the encoded codeword(s) with an interference level and a parity check syndrome at each of the plurality of time steps, using the training samples to train a neural network based decoder to decode codewords encoded using an error correction code by (1) estimating a multiplicative interference included in the encoded codeword(s) based on reverse diffusion applied to the encoded codeword(s) across the time steps, (2) computing an additive interference included in the encoded codewords based on the multiplicative interference, and (3) recovering the codeword(s) by removing the additive interference.

Abstract: In one embodiment, a method includes receiving a first image depicting a context including one or more persons having one or more respective poses, receiving a second image depicting a target person having an original pose, where the target person is to be inserted into the context depicted in the first image, generating a target segmentation mask specifying a new pose for the target person in the context of the first image based on the first image, generating a third image depicting the target person having the new pose based on the second image and the target segmentation mask, and generating an output image based on the first image and the third image, the output image depicting the one or more persons having the one or more respective poses and the target person having the new pose.

Abstract: A technique and system for counting the number of repetitions of approximately the same action in an input video sequence using 3D convolutional neural networks is disclosed. The proposed system runs online and not on the complete video. It analyzes sequentially blocks of 20 non-consecutive frames. The cycle length within each block is evaluated using a deep network architecture and the information is then integrated over time. A unique property of the disclosed method is that it is shown to successfully train on entirely synthetic data, created by synthesizing moving random patches. It therefore effectively exploits the high generalization capability of deep neural networks. Coupled with a region of interest detection mechanism and a suitable mechanism to identify the time scale of the video, the system is robust enough to handle real world videos collected from youtube and elsewhere, as well as non-video signals such as sensor data revealing repetitious physical movement.

Abstract: A video generation system is described that extracts one or more characters or other objects from a video, re-animates the character, and generates a new video in which the extracted characters. The system enables the extracted character(s) to be positioned and controlled within a new background scene different from the original background scene of the source video. In one example, the video generation system comprises a pose prediction neural network having a pose model trained with (i) a set of character pose training images extracted from an input video of the character and (ii) a simulated motion control signal generated from the input video. In operation, the pose prediction neural network generates, in response to a motion control input from a user, a sequence of images representing poses of a character. A frame generation neural network generates output video frames that render the character within a scene.

Abstract: A neural network system for detecting at least one object in at least one image, the system includes a plurality of object detectors. Each object detector receives respective image information thereto. Each object detector includes a respective neural network. Each the neural network including a plurality of layers. Layers in different object detectors are common layers when the layers receive the same input thereto and produce the same output therefrom. Common layers are computed only once during object detection for all the different object detectors.

Abstract: Disclosed herein a system, a method and a device for generating a voice model for a user. A device can include an encoder and a decoder to generate a voice model for converting text to an audio output that resembles a voice of the person sending respective text. The encoder can includes a neural network and can receive a plurality of audio samples from a user. The encoder can generate a sequence of values and provide the sequence of values to the decoder. The decoder can establish, using the sequence of values and one or more speaker embeddings of the user, a voice model corresponding to the plurality of audio samples of the user.

Abstract: In one embodiment, a method includes generating a keypoint pose and a dense pose for a first person in a first pose based on a first image comprising the first person in the first pose, generating an input semantic segmentation map corresponding to a second person in a second pose based on a second image comprising the second person in the second pose, generating a target semantic segmentation map corresponding to the second person in the first pose by processing the keypoint pose, the dense pose, and the input segmentation map using a first machine-learning model, generating an encoding vector representing the second person based on the second image, and generating a target image of the second person in the first pose by processing the encoding vector and the target segmentation map using a second machine-learning model.

Abstract: In one embodiment, a method includes generating a first identity encoding representing a first facial identity of the person based on an image of a person, generating a second identity encoding representing a second facial identity different from the first facial identity of the person based on the first identity encoding, generating a source encoding by using an encoder to process a source image of the person having an expression, generating an intermediate image by using a decoder to process the source encoding and the second identity encoding, the intermediate image including a face having the second facial identity and the expression of the person in the source image, and generating an output image by blending the source image with facial features of the face in the intermediate image.

Abstract: In one embodiment, a method includes receiving a mixed audio signal comprising a mixture of voice signals associated with a plurality of speakers, generating first audio signals by processing the mixed audio signal using a first machine-learning model configured with a first number of output channels, determining that at least one of the first number of output channels is silent based on the first audio signals, generating second audio signals by processing the mixed audio signal using a second machine-learning model configured with a second number of output channels that is fewer than the first number of output channels, determining that each of the second number of output channels is non-silent based on the second audio signals, and using the second machine-learning model to separate additional mixed audio signals associated with the plurality of speakers.

Abstract: In one embodiment, a method includes inputting an encoded message with noise to a neural-networks model comprising a variable and a check layers of nodes, each node being associated with at least one weight and a hyper-network node, updating the weights associated with the variable layer of nodes by processing the encoded message using the hyper-network nodes associated with the variable layer of nodes, generating a first set of outputs by processing the encoded message using the variable layer of nodes and their respective updated weights, updating the weights associated with the check layer of nodes by processing the first set of outputs using the hyper-network nodes associated with the check layer of nodes, and generating a decoded message without noise using the neural-networks model by using at least the first set of outputs and the check layer of nodes and their respective updated weights.

Abstract: A technique and system for counting the number of repetitions of approximately the same action in an input video sequence using 3D convolutional neural networks is disclosed. The proposed system runs online and not on the complete video. It analyzes sequentially blocks of 20 non-consecutive frames. The cycle length within each block is evaluated using a deep network architecture and the information is then integrated over time. A unique property of the disclosed method is that it is shown to successfully train on entirely synthetic data, created by synthesizing moving random patches. It therefore effectively exploits the high generalization capability of deep neural networks. Coupled with a region of interest detection mechanism and a suitable mechanism to identify the time scale of the video, the system is robust enough to handle real world videos collected from youtube and elsewhere, as well as non-video signals such as sensor data revealing repetitious physical movement.

Abstract: A video generation system is described that extracts one or more characters or other objects from a video, re-animates the character, and generates a new video in which the extracted characters. The system enables the extracted character(s) to be positioned and controlled within a new background scene different from the original background scene of the source video. In one example, the video generation system comprises a pose prediction neural network having a pose model trained with (i) a set of character pose training images extracted from an input video of the character and (ii) a simulated motion control signal generated from the input video. In operation, the pose prediction neural network generates, in response to a motion control input from a user, a sequence of images representing poses of a character. A frame generation neural network generates output video frames that render the character within a scene.

Abstract: Disclosed herein a system, a method and a device for generating a voice model for a user. A device can include an encoder and a decoder to generate a voice model for converting text to an audio output that resembles a voice of the person sending respective text. The encoder can includes a neural network and can receive a plurality of audio samples from a user. The encoder can generate a sequence of values and provide the sequence of values to the decoder. The decoder can establish, using the sequence of values and one or more speaker embeddings of the user, a voice model corresponding to the plurality of audio samples of the user.

Abstract: A technique and system for counting the number of repetitions of approximately the same action in an input video sequence using 3D convolutional neural networks is disclosed. The proposed system runs online and not on the complete video. It analyzes sequentially blocks of 20 non-consecutive frames. The cycle length within each block is evaluated using a deep network architecture and the information is then integrated over time. A unique property of the disclosed method is that it is shown to successfully train on entirely synthetic data, created by synthesizing moving random patches. It therefore effectively exploits the high generalization capability of deep neural networks. Coupled with a region of interest detection mechanism and a suitable mechanism to identify the time scale of the video, the system is robust enough to handle real world videos collected from youtube and elsewhere, as well as non-video signals such as sensor data revealing repetitious physical movement.