Abstract: The label, method, and system for automated recognition of products may be used at customer self-service checkouts, and for sorting products in automated warehouses. The label includes a base and a graphic code applied to the base that matches the ID of the product. The graphic code contains at least four areas located in the different parts of the label. Each area encodes a part of the ID and includes at least two graphic elements, one encoding at least one character of the ID, and the other encoding the position of such character in the ID. The product recognition system includes at least one scanning device and a data processing unit. The method of recognition includes scanning the label, receiving pictures or images of areas of the graphic code, recognizing the graphic elements of the code, decoding data, and determining the ID to identify the product.

Abstract: Based upon the principles of randomness and self-modification a novel computing machine is constructed. This computing machine executes computations, so that it is difficult to apprehend by an adversary and hijack with malware. These methods can also be used to help thwart reverse engineering of proprietary algorithms, hardware design and other areas of intellectual property. Using quantum randomness in the random instructions and self-modification in the meta instructions, creates computations that are incomputable by a digital computer. In an embodiment, a more powerful computational procedure is created than a computational procedure equivalent to a digital computer procedure. Current digital computer algorithms and procedures can be constructed or designed with ex-machine programs, that are specified by standard instructions, random instructions and meta instructions. A novel computer is invented so that a program's execution is difficult to apprehend.

Abstract: While a qubit control system (e.g., a laser system) is in a first configuration, it causes a qubit state (as represented as a point on the surface of a Bloch sphere) of a quantum state carrier (QSC), e.g., an atom, to rotate in a first direction from an initial qubit state to a first configuration qubit state. While the qubit control system is in a second configuration, it causes the QSC state to rotate in a second direction opposite the first direction from the first configuration qubit state to a second configuration qubit state. The second configuration qubit state is read out as a |0 or |1. Repeating these actions results in a distribution of |0s and |1s that can be used to determine which of the two configurations results in higher Rabi frequencies. Iterating the above for other pairs of configurations can identify a configuration that delivers the most power to the QSC and thus yields the highest Rabi frequency.

Abstract: Provided is a system, an information processing method, and a non-transitory storage medium that hardly cause improper operations when a plurality of quantum processors is connected to configure a logical quantum bit.

Abstract: Methods and systems for a reinforcement learning system. A spatial and temporal representation of an observed state of an environment is encoded. A previous state is estimated from a given state and a size of a reward is adjusted based on a difference between the estimated previous state and the previous state.

Abstract: Systems, methods, and non-transitory computer-readable media can be configured to determine a targeted scenario and a mission associated with the targeted scenario. A route to a location associated with the mission can be determined based at least in part on a likelihood of encountering the targeted scenario, wherein the likelihood of encountering the targeted scenario is based at least in part on a frequency with which scenarios similar to the targeted scenario were encountered at the location. Whether the targeted scenario was encountered can be determined based on an evaluation of captured sensor data associated with the mission upon passing the location.

Abstract: A request is received associated with a review. Within first content, a first field of interest and a second field of interest are identified and within second content, a third field of interest and a fourth field of interest are identified. A review is generated that includes a first indication of the first field of interest and a second indication of the second field of interest within the first content, as well as a third indication of the third field of interest and a fourth indication of the fourth field of interest within the second content. The review is transmitted to a device of a reviewer for reviewing the content.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for transformation for machine learning pre-processing. In some implementations, an instruction to create a model is obtained. A determination is made whether the instruction specifies a transform. In response to determining that the instruction specifies a transform, a determination is made as to whether the transform requires statistics on the training data. The training data is accessed. In response to determining that the transform requires statistics on the training data, transformed training data is generated from both the training data and the statistics. A model is generated with the transformed training data. A representation of the transform and the statistics is stored as metadata for the model.

Abstract: A system monitoring a plurality of models generated through machine learning includes a monitoring unit that perform warning of a specific item of an input corresponding to a predetermined condition if a prediction result by a first model using the input including a plurality of values satisfies the predetermined condition, and a provision unit that provides a message prompting setting of a condition which is a monitoring target by the monitoring unit in the specific item with regard to a second model different from the first model. The provision unit provides the message in at least one of a case in which the predetermined condition is set for the first model, a case in which the prediction result by the first model is determined to satisfy the predetermined condition, and a case in which the second model is registered in the system.

Abstract: A system for customizing informed advisor pairings, the system including a computing device. The computing device is configured to identify a user feature wherein the user feature contains a user biological extraction. The computing device is configured to generate using element training data and using a first machine-learning algorithm a first machine-learning model that outputs advisor elements. The computing device receives an informed advisor element relating to an informed advisor. The computing device determines using output advisor elements whether an informed advisor is compatible for a user.

Abstract: A system and method include input of data records to a first trained predictive model to obtain a predicted value associated with each input data record. A model region is then associated with each of the input data records based on the first trained predictive model, the input data records and the predicted values. Enhanced input data records are generated by, for each model region, adding derived values of engineered features associated with the model region to input data records associated with the model region and default values of the engineered features associated with the model region to input training records not associated with the model region. The enhanced input data records are input to a second trained predictive model to obtain an enhanced predicted value associated with each input data record.

Abstract: The present application provides a model training, image processing method, device, storage medium, and program product relating to deep learning technology, which are able to screen auxiliary image data with image data for learning a target task, and further fuse the target image data and the auxiliary image data, so as to train a built and to-be-trained model with the fusion-processed fused image data. This implementation can increase the amount of data for training the model, and the data for training the model is determined is based on the target image data, which is suitable for learning the target task. Therefore, the solution provided by the present application can train an accurate target model even if the amount of target image data is not sufficient.

Abstract: Systems are provided for facilitating the building and use of models used to make data preparation recommendations. The systems identify ground truth from a plurality of notebooks and utilizes the ground truth to generate the corresponding data preparation recommendation models. The data preparation recommendation models are used to predict accurate (e.g., useful and relevant) data preparations steps based on user input and user notebook data. The data preparation computing system generates a recommendation prompt based on output from the data preparation recommendation model that can be viewed and/or selected by the user to be applied to the user's notebook data.

Abstract: Methods and systems for building and maintaining model(s) of a physical process are disclosed. One method includes receiving training data associated with a plurality of different data sources, and performing a clustering process to form one or more clusters. For each of the one or more clusters, the method includes building a data model based on the training data associated with the data sources in the cluster, automatically performing a data cleansing process on operational data based on the data model, and automatically updating the data model based on updated training data that is received as operational data. For data sources excluded from the clusters, automatic building, data cleansing, and updating of models can also be applied.

Abstract: There is provided a system and method for compression and decompression of a data stream used by machine learning networks. The method including: encoding each value in the data stream, including: determining a mapping to one of a plurality of non-overlapping ranges, each value encoded as a symbol representative of the range and a corresponding offset; and arithmetically coding the symbol using a probability count; storing a compressed data stream including the arithmetically coded symbols and the corresponding offsets; and decoding the compressed data stream with arithmetic decoding using the probability count, the arithmetic decoded symbols use the offset bits to arrive at a decoded data stream; and communicating the decoded data stream for use by the machine learning networks.

Abstract: Methods, systems and computer program products are described to improve machine learning (ML) model-based classification of data items by identifying and removing inaccurate training data. Inaccurate training samples may be identified, for example, based on excessive variance in vector space between a training sample and a mean of category training samples, and based on a variance between an assigned category and a predicted category for a training sample. Suspect or erroneous samples may be selectively removed based on, for example, vector space variance and/or prediction confidence level. As a result, ML model accuracy may be improved by training on a more accurate revised training set. ML model accuracy may (e.g., also) be improved, for example, by identifying and removing suspect categories with excessive (e.g., weighted) vector space variance. Suspect categories may be retained or revised. Users may (e.g., also) specify a prediction confidence level and/or coverage (e.g., to control accuracy).

Abstract: Certain aspects of the present disclosure provide techniques for managing the transmission of mixed-modality messages using machine learning models. An example method generally includes generating, using a first machine learning model, an embedding representation of a mixed-modality message. The mixed-modality message is classified as an effective message or an ineffective message using a second machine learning model and the embedding representation of the mixed-modality message. One or more actions are taken to manage transmission of the mixed-modality message based on the classifying the mixed-modality message as an effective message or an ineffective message.

Abstract: Certain aspects of the present disclosure provide techniques for orchestrating a user experience using natural language input. A user experience is orchestrated within an ecosystem of features for which a plurality of corresponding tokens is defined. Natural language describing a desired user experience result is received by a user experience orchestrator. A sequence of tokens corresponding to operations belonging to an ecosystem of features which produce a correct result for the natural language input can be identified by a trained large language model and executed by the user experience orchestrator using a token operator. The output operations determined by the model to produce or be likely to produce the correct result based on the natural language input can be disambiguated, confirmed, and/or executed.

Abstract: A system for alimentary instruction sets derived from artificial intelligence systems for vibrant constitutional guidance, as derived using one or more machine-learning procedures from training data relating prognostic and ameliorative labels. A physical performance instruction set is derived from the alimentary instruction sets using one or more physical performance entity profiles.

Abstract: Provided is a method for training distributed machine learning models. The method may include initializing a distributed machine learning model on a plurality of computing devices. Training data associated with a plurality of samples may be received. Each sample may be forward propagated through the distributed machine learning model to generate an output. A loss for each sample of the plurality of samples may be determined based on the output. The loss for each sample may be backward propagated to each computing device. The parameter(s) of each computational node may be asynchronously updated based on the loss as it is backward propagated and/or while at least one of the samples is forward propagating. The parameter(s) may be stored and/or communicated to the other computing devices. Each of the other computing devices of the plurality of computing devices may store the parameter(s). A system and computer program product are also disclosed.

Abstract: A method includes receiving information associated with a requested operator. The method further includes, in response to receiving the information, generating, by a processing device executing a machine learning model, an artificial intelligence (AI)-based solution to the requested operator, wherein the AI-based solution comprises a plurality of machine-learning models. The method further includes displaying an option to access the AI-based solution in a marketplace platform. The method further includes receiving information associated with a requested operator, and generating, by a processing device executing a first machine learning model, a skeleton architecture of an AI-based solution to the operator based on the information.

Abstract: A computer system has a first machine learning module configured to predict a probability of a respective option being selected by a particular user if presented to that user via a computer app. A second machine learning module is configured to determine a respective confidence value associated with the probability. A third module uses the predicted probabilities and confidence values to determine at least one option to be presented to the particular user.

Abstract: The present disclosure is directed to an automated neural architecture search approach for designing new neural network architectures such as, for example, resource-constrained mobile CNN models. In particular, the present disclosure provides systems and methods to perform neural architecture search using a novel factorized hierarchical search space that permits layer diversity throughout the network, thereby striking the right balance between flexibility and search space size. The resulting neural architectures are able to be run relatively faster and using relatively fewer computing resources (e.g., less processing power, less memory usage, less power consumption, etc.), all while remaining competitive with or even exceeding the performance (e.g., accuracy) of current state-of-the-art mobile-optimized models.

Abstract: An activation function processing method includes processing a first activation function in a first mode by referring to a shared lookup table that includes a plurality of function values of the first activation function; and processing a second activation function in a second mode by referring to the shared lookup table, the second activation function being a different function than the first activation function.

Abstract: The present disclosure describes an artificial neural network circuit including: at least one crossbar circuit to transmit a signal between layered neurons of an artificial neural network, the crossbar circuit including multiple input bars, multiple output bars arranged intersecting the input bars, and multiple memristors that are disposed at respective intersections of the input bars and the output bars to give a weight to the signal to be transmitted; a processing circuit to calculate a sum of signals flowing into each of the output bars while a weight to a corresponding signal is given by each of the memristors; a temperature sensor to detect environmental temperature; and an update portion that updates a trained value used in the crossbar circuit and/or the processing circuit.

Abstract: A parallel convolutional neural network is provided. The CNN is implemented by a plurality of convolutional neural networks each on a respective processing node. Each CNN has a plurality of layers. A subset of the layers are interconnected between processing nodes such that activations are fed forward across nodes. The remaining subset is not so interconnected.

Abstract: A method of processing of a sparsity-aware neural processing unit includes receiving a plurality of input activations (IA); obtaining a weight having a non-zero value in each weight output channel; storing the weight and the IA in a memory, and obtaining an input channel index comprising a memory address location in which the weight and the IA are stored; and arranging the non-zero weight of each weight output channel according to a row size of an index matching unit (IMU) and matching the IA to the weight in the IMU comprising a buffer memory storing the input channel index.

Abstract: A synapse string includes first and second cell strings each having a plurality of memory cell elements connected in series and first switch elements connected to first or second ends of the first and second cell strings, respectively. The memory cell elements of the first cell string and the memory cell elements of the second cell string are in a one-to-one correspondence, and a pair of the memory cell elements being in a one-to-one correspondence has terminals to which a read voltage is applied connected to each other to constitute one synapse morphic element, so that the synapse string includes a plurality of synapse morphic elements connected in series. A synapse string array architecture enables forward propagation and backward propagation by implementing high-density synapse strings, so that the synapse string array architecture can be applied to a neural network capable of inferencing and on-chip learning, along with inference and recognition.

Abstract: Methods, systems, and apparatus, including computer-readable media, are described for interleaving memory requests to accelerate memory accesses at a hardware circuit configured to implement a neural network model. A system generates multiple requests that are processed against a memory of the system. Each request is used to retrieve data from the memory. For each request, the system generates multiple sub-requests based on a respective size of the data to be retrieved using the request. The system generates a sequence of interleaved sub-requests that includes respective sub-requests of a first request interleaved among respective sub-requests of a second request. Based on the sequence of interleaved sub-requests, a module of the system receives respective portions of data accessed from different address locations of the memory. The system processes each of the respective portions of data to generate a neural network inference using the neural network model implemented at the hardware circuit.

Abstract: A method of compressing kernels comprising detecting a plurality of replicated kernels. The plurality of replicated kernels comprise kernels. The method also comprises generating a composite kernel from the replicated kernels. The composite kernel comprises kernel data and meta data indicative of the rotations applied to the composite kernel data. The method also comprises storing a composite kernel.

Abstract: Embodiments of the invention provide a system, media, and method for deep learning applications in physical design verification. Generally, the approach includes maintaining a pattern library for use in training machine learning model(s). The pattern library being generated adaptively and supplemented with new patterns after review of new patterns. In some embodiments, multiple types of information may be included in the pattern library, including validation data, and parameter and anchoring data used to generate the patterns. In some embodiments, the machine learning processes are combined with traditional design rule analysis. The patterns being generated and adapted using a lossless process that encodes the information of a corresponding area of a circuit layout.

Abstract: Techniques for generating a set of Deep Learning (DL) models are described. An example method includes training an initial set of DL models using the training data, wherein a topology of each of the DL models is determined based on the parameters vector. The method also includes generating a set of estimate performance functions for each of the DL models in the initial set based on the set of edge-related metrics, and generating a plurality of objective functions based on the set of estimated performance functions. The method also includes generating a final DL model set based on the objective functions, receiving a user selection of a selected DL model from the final DL model set, and deploying the selected DL model to an edge device.

Abstract: Methods, systems, and devices for distributed hyperparameter tuning and load balancing are described. A device (e.g., an application server) may generate a first set of combinations of hyperparameter values associated with training a mathematical model. The mathematical model may include a machine learning model, an optimization model, or any combination. The device may identify a subset of combinations from the first set of combinations that are associated with a computational runtime that exceeds a first threshold and may distribute the subset of combinations across a set of machines. The device may then test each of the first set of combinations in a parallel processing operation to generate a first set of validation error values and may test a second set of combinations of hyperparameter values using an objective function that is based on the first set of validation error values.

Abstract: Described is a system for training a neural network for estimating surface normals for use in operating an autonomous platform. The system uses a parallelizable k-nearest neighbor sorting algorithm to provide a patch of points, sampled from the point cloud data, as input to the neural network model. The points are transformed from Euclidean coordinates in a Euclidean space to spherical coordinates. A polar angle of a surface normal of the point cloud data is estimated in the spherical coordinates. The trained neural network model is utilized on the autonomous platform, and the estimate of the polar angle of the surface normal is used to guide operation of the autonomous platform within the environment.

Abstract: Methods, systems, and apparatus for designing a quantum control trajectory for implementing a quantum gate using quantum hardware. In one aspect, a method includes the actions of representing the quantum gate as a sequence of control actions and applying a reinforcement learning model to iteratively adjust each control action in the sequence of control actions to determine a quantum control trajectory that implements the quantum gate and reduces leakage, infidelity and total runtime of the quantum gate to improve its robustness of performance against control noise during the iterative adjustments.

Abstract: Techniques and apparatuses are described for enabling base station-user equipment messaging regarding deep neural networks. A network entity (base station 121, core network server 320) determines a neural network formation configuration (architecture and/or parameter configurations 1208) for a deep neural network (deep neural network(s) 604, 608, 612, 616) for processing communications transmitted over the wireless communication system. The network entity (base station 121, core network server 302) communicates the neural network formation configuration to a user equipment (UE 110). The user equipment (UE 110) configures a first neural network (deep neural network(s) 608, 612) based on the neural network formation configuration. In implementations, the user equipment (UE 110) recovers information communicated over the wireless network using the first neural network (deep neural network(s) 608, 612).

Abstract: Disclosed is an in-memory device for operation of a multi-bit weight. A multi-bit memory cell array according to an exemplary embodiment of the present invention includes at least one multi-bit unit which stores input data based on an input signal and outputs a per-group sum value summed for every group by applying a multi-bit weight to the stored input data; and a final summation unit which is connected to at least one multi-bit unit, adjusts a ratio for every group to receive the peer-group sum value, and outputs a final output value by summing the input per-group sum value.

Abstract: Disclosed herein is an image preprocessing/analysis apparatus using machine learning-based artificial intelligence. The image preprocessing apparatus includes a computing system, and the computing system includes: a processor; a communication interface configured to receive an input image; and an artificial neural network configured to generate first and second preprocessing conditions through inference on the input image. The processor includes a first preprocessing module configured to generate a first preprocessed image and a second preprocessing module configured to generate a second preprocessed image.

Abstract: A device includes a state machine. The state machine includes a plurality of blocks, where each of the blocks includes a plurality of rows. Each of these rows includes a plurality of programmable elements. Furthermore, each of the programmable elements are configured to analyze at least a portion of a data stream and to selectively output a result of the analysis. Each of the plurality of blocks also has corresponding block activation logic configured to dynamically power-up the block.

Abstract: An information processing apparatus and an information processing method capable of accurately recognizing an object to be sensed are provided. One or a plurality of learning models are selected from among learning models corresponding to a plurality of categories, a priority of each of the selected learning models is set, observation data obtained by sequentially compounding pieces of sensor data applied from a sensor is analyzed using the learning model and the priority of the learning model, a setting of sensing at a next cycle is selected based on an analysis result, predetermined control processing is executed in such a manner that the sensor performs sensing at the selected setting of sensing, and the learning model corresponding to the category estimated as the category to which a current object to be sensed belongs with a highest probability is selected based on the categories to each of which the previously recognized object to be sensed belongs.

Abstract: Methods, devices and systems for training a pattern recognition system are described. In one example, a method for training a sign language translation system includes generating a three-dimensional (3D) scene that includes a 3D model simulating a gesture that represents a letter, a word, or a phrase in a sign language. The method includes obtaining a value indicative of a total number of training images to be generated, using the value indicative of the total number of training images to determine a plurality of variations of the 3D scene for generating of the training images, applying each of plurality of variations to the 3D scene to produce a plurality of modified 3D scenes, and capturing an image of each of the plurality of modified 3D scenes to form the training images for a neural network of the sign language translation system.

Abstract: Among a great deal of other disclosure and scope, systems and methods are enclosed that enable for highly efficient labeling of data. For example, in some of many cases, a novel methodology for ranking vectors most useful to label next is disclosed. In such an example, a neural network is trained to predict this ranking methodology upon being given a set of heuristics from which to assess the given problem space. A user can continue the cycle of identifying a set of candidate vectors to label, compiling relevant heuristics from said vectors, ranking vectors via the trained neural network, selecting a subset of the ranked vectors, inquiring an oracle regarding the true labels of the vectors, and then appending the subset of newly labelled vectors to the labelled set of vectors until satisfaction.

Abstract: Training images can be synthesized in order to obtain enough data to train a model (e.g., a neural network) to recognize various classifications of a type of object. Images can be synthesized by blending images of objects labeled using those classifications into selected background images. To improve results, one or more operations are performed to determine whether the synthesized images can still be used as training data, such as by verifying one or more objects of interested represented in those images is not occluded, or at least satisfies a threshold level of acceptance. The training images can be used with real world images to train the model.

Abstract: A method of processing data for a deep learning system driven by a plurality of heterogeneous resources is provided. The method includes, when a first task including at least one of a plurality of operations is to be performed, receiving first path information indicating a first computing path for the first task. The first computing path includes a sequence of operations included in the first task and a driving sequence of resources for performing the operations included in the first task. The method further includes setting data representation formats of the resources for performing the operations included in the first task based on data representation information and the first path information. The data representation information indicates an optimized data representation format for each of the plurality of heterogeneous resources.

Abstract: Various techniques are described for platform management of integrated access of public and privately-accessible datasets utilizing federated query generation and query schema rewriting optimization, including receiving at a dataset access platform a query formatted according to a first data schema, generating a copy of the query, saving the query and the copy to a datastore, parsing the copy of the query in the first schema using an inference engine, determining whether the query comprises data associated with an access control condition associated with accessing the dataset, the access control condition being configured to indicate whether the query is permitted to access the dataset, and rewriting, using a proxy server, the copy of the query in a second schema by converting the copy of the query into a triple associated with the query and another triple associated with the access control condition.

Abstract: There is described a computer-implemented method and system for classifying images, the computer-implemented method comprising: receiving an image to be classified, generating a vector representation of the image to be classified using an image embedding method, comparing the vector representation of the image to predefined vector representations of the predefined image categories, and identifying a relevant category amongst the predefined image categories based on the comparison, the relevant category being associated with the image to be classified and outputting the relevant category.

Abstract: The present disclosure discloses a system and method for distributed neural network training. The method includes: computing, by a plurality of heterogeneous computation units (HCUs) in a neural network processing system, a first plurality of gradients from a first plurality of samples; aggregating the first plurality of gradients to generate an aggregated gradient; computing, by the plurality of HCUs, a second plurality of gradients from a second plurality of samples; aggregating, at each of the plurality of HCUs, the aggregated gradient with a corresponding gradient of the second plurality of gradients to generate a local gradient update; and updating, at each of the plurality of HCUs, a local copy of a neural network with the local gradient update.

Abstract: A method and device for model compression of a neural network. The method comprises: recording input and output parameters of each layer of network in a network structure; dividing the network structure into several small networks according to the input and output parameters; setting a pruning flag bit of a first convolutional layer in each small network to be zero to obtain a pruned small network; training each pruned small network to obtain a network weight and a weight mask; recording a pruned channel index number of each convolutional layer of a pruned small network with the weight mask of zero; and carrying out decomposition calculation on each pruned small network according to the pruned channel index number. According to the method, a calculation amount and the size of a model is reduced, and during network deployment, the model can be loaded with one click, thus reducing usage difficulty.

Abstract: A method for sequence-to-sequence prediction using a neural network model includes generating an encoded representation based on an input sequence using an encoder of the neural network model and predicting an output sequence based on the encoded representation using a decoder of the neural network model. The neural network model includes a plurality of model parameters learned according to a machine learning process. At least one of the encoder or the decoder includes a branched attention layer. Each branch of the branched attention layer includes an interdependent scaling node configured to scale an intermediate representation of the branch by a learned scaling parameter. The learned scaling parameter depends on one or more other learned scaling parameters of one or more other interdependent scaling nodes of one or more other branches of the branched attention layer.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for neural network compression. In one aspect, a method comprises receiving a neural network and identifying a particular set of multiple weights of the neural network. Multiple anchor points are determined based on current values of the particular set of weights of the neural network. The neural network is trained by, at each of multiple training iterations, performing operations comprising adjusting the values of the particular set of weights by backpropagating gradients of a loss function. The loss function comprises a first loss function term based on a prediction accuracy of the neural network and a second loss function term based on a similarity of the current values of the particular set of weights to the anchor points. After training, the values of the particular set of weights are quantized based on the anchor points.