Abstract: Systems, apparatuses and methods provide technology that identifies a plurality of data samples of a data set. In response to a trigger, the technology identifies a first subset of samples of the plurality of data samples that are outside a range associated with a decision boundary of a machine learning model, and removes the first subset of samples from the data set to generate a modified training set. The technology train the machine learning model based on the modified training set.

Abstract: A display method includes an electronic device that obtains preset information, where the preset information includes information about an application installed on the electronic device, and the information includes at least one of a name, an icon, and a package name of the application. The electronic device generates a first window based on the preset information, where the first window includes one of a plurality of desktops of the electronic device. The electronic device sends the first window to the large-screen device to enable the large-screen device to display a first user interface, where the first user interface includes the first window.

Abstract: Systems, methods, and computer-readable media are described for training and implementing logic gate networks for inference tasks. An untrained node network includes a first set of nodes parameterized by differentiable parameters, each associated with a predefined set of potential logic gate operators, and a second set of nodes corresponding to predefined logic gate operators. During training, node outputs of the first set are computed using a first relaxation, and node outputs of the second set are computed using a second relaxation different from the first relaxation. Updated differentiable parameters are applied over multiple training iterations, and a fixed logic-gate network is generated by selecting logic gate operators for at least some nodes in the first set while retaining the predefined logic gate operators for the second set. In some embodiments, the second relaxation imposes structural inductive bias. ASICs and FPGAs implementing fixed logic gate networks are also described.

Abstract: A system for facilitating user interaction with interactive areas includes a memory encoding processor-executable routines. The system also includes a processor configured to access the memory and to execute the processor-executable routines. The processor may identify data a user of an interactive area based on identifying data obtained at the interactive area. The processor may also utilize a trained machine learning model personalized for the user, wherein the trained machine learning model personalized for the user is configured to recognize idiosyncrasies of the user. The processor may also utilize the trained machine learning model personalized for the user in detecting an idiosyncratic task performed by the user interacting with the interactive area to activate a special effect associated with the interactive area based on interactive data obtained at the interactive area. The processor may further instruct the initiation of the special effect in response to detecting the idiosyncratic task.

Abstract: Disclosed is an operating method of a computing device. The operating method of a computing device includes generating quantum mapping basis information based on a fault-tolerant constraint, generating quantum circuits and initial qubit mappings by performing a quantum circuit mapping as much as the number of times based on the fault-tolerant constraint, the quantum mapping basis information, and different random initial qubit mappings, and selecting one quantum circuit and one initial qubit mapping from among the quantum circuits and the initial qubit mappings, respectively.

Abstract: Mechanisms are provided for automatic identification of a reconciliation computer tool for producing coherent reconciled data from base data generated by a computer model. A machine learning training operation is executed on one or more performance prediction computer model(s) (PPCMs) based on first input features of at least one hierarchical dataset, and second input features of a plurality of different reconciliation computer tools. The PPCM(s) generate a prediction of performance of a corresponding reconciliation computer tool based on the first and second input features. Features are extracted from a runtime hierarchical dataset and input into the trained PPCM(s) which generate predictions of performance of a plurality of reconciliation computer tools based on the extracted features of the runtime hierarchical dataset. The reconciliation computer tools are ranked relative to one another based on the predictions of performance.

Abstract: A processor-implemented method with neural network processing includes: determining whether a portion of a population comprising a plurality of instances to which different mixed-precision quantizations are applied for a neural network satisfies convergence criteria; generating, in response to the determination that the portion satisfies the convergence criteria, a new instance using the portion; and updating the population by adding the new instance to the population.

Abstract: In an embodiment, a method includes receiving input data. The method also includes collecting data attributes from the input data based on a type of the input data. The method also includes classifying the input data using the data attributes. The method also includes performing class-imbalance removal on the classified input data, the performing yielding a balanced dataset. The method also includes creating integrated data using the balanced dataset, the integrated data including the balanced dataset integrated with data-attribute intelligence. The method also includes representing at least a portion of the integrated data as structured knowledge for execution of a particular task. The method also includes applying a plurality of extrapolation algorithms to the structured knowledge to yield raw augmented data. The method also includes generating an extrapolated dataset using the raw augmented data. The method also includes constructing a synthetic dataset based on the extrapolated dataset.

Abstract: According to an embodiment, a point process learning method executed by a computer includes: an input procedure of inputting a learning data set including at least first event data representing a series of occurrences of first events; a division procedure of dividing the first event data included in the learning data set by using a prediction time observation area including at least a time series when predicting future event occurrence; and a learning procedure of learning a model parameter including a parameter of an intensity function of a predetermined point process model by using a divided learning data set divided in the division procedure.

Abstract: A method, system, and computer program product are configured to: receive an input machine learning model being developed for deployment in a production environment; create a feature vector based on metadata of the input machine learning model and a data source used by the input machine learning model; determine a risk tier of the input machine learning model by classifying the input machine learning model into one of plural predefined tiers using the feature vector with a model-tiering machine learning model; and provide validation information to an evaluator wherein the validation information is based on the risk tier and is used to validate the input machine learning model in accordance with the risk tier.

Abstract: A method includes in a first iteration receiving a machine learning (ML) network model comprising a plurality of ML operations in high-level code; generating an internal representation (IR) for the ML network model, the IR mapped to one or more components in a multi-processing tile device; generating primitive functions based on the IR; generating an allocation list based on the primitive functions; determining when a tensor data within the allocation list is no longer needed; and inserting a deallocation function associated with the tensor data to the primitive functions to form an updated primitive functions, the inserting frees up a memory space associated with the tensor data when the tensor is no longer needed; and in a second iteration generating a compilation of the updated primitive functions to map the IR to the one or more components in the multi-processing device, wherein the compilation is a low-level instructions.

Abstract: In example embodiments, techniques are provided for customizing a ML model for a specific user absent user-coding. The techniques may provide a “black box” service to the user where the intricacies of ML model training are abstracted, and the user simply provides source data and makes high level selections. The techniques may be used with a variety of types of ML model architectures and ML Pipelines.

Abstract: The present disclosure relates to a method and an apparatus for federated learning of an artificial intelligence model. According to an exemplary embodiment of the present disclosure, a federated learning method of an artificial intelligence model includes: training a first local artificial intelligence model and a second local artificial intelligence model using data sets of a first client and a second client among the plurality of clients; calculating performance values for the first local artificial intelligence model and the second local artificial intelligence model by transmitting the first local artificial intelligence model to the second client and transmitting the second local artificial intelligence model to the first client; comparing the performance values to remove one of the first client and the second client; and training a global model using a client which is not removed.

Abstract: Modality agnostic Large Codeword Model (“LCM”) is an advanced deep learning architecture that processes discrete, compressed data representations called codewords across multiple modalities. Unlike traditional models using raw tokens and dense embeddings, LCMs efficiently handle diverse input types including text, images, audio, and video. The system employs a modality agnostic encoder, unified codebook, and multimodal machine learning core to capture inherent data structures and patterns. This approach enables more generalizable and interpretable feature learning, facilitating transfer learning across domains. The LCM's scalable and flexible architecture includes components for modality-specific processing, cross-modal attention, and joint representation learning. With its computational efficiency and versatility, the Modality Agnostic LCM offers significant potential for various AI applications, including natural language processing, computer vision, and multimodal reasoning.

Abstract: An approach is provided that trains an artificial intelligence (AI) system, such as a neural network, to process IT ticket data. The approach receives IT tickets from various ticket sources. Ticket vectors corresponding to each of the IT tickets are computed. An analysis is performed using the ticket vectors and a node vector that corresponds to a network topology. The analysis is performed using a corpus of IT ticket data. An IT ticket model used by the AI system is trained based on the analysis. Responses are provided to requestors of the AI system using the trained IT ticket model.

Abstract: Embodiments for providing expert-in-the-loop training of machine learning models in a computing environment by a processor. A performance of a machine learning model may be learned. Feedback for the machine learning model may be received based on learning the performance the machine learning model, where the feedback includes domain knowledge provided by a domain expert. The machine learning model may be trained or updated based the feedback of the performance of the machine learning model.

Abstract: Systems and methods of the present disclosure enable identifying labelling a source signal data signature using a computing system to test candidate chain oracle models by iteratively performing, for each particular number of neural network models in the range of the number of neural network models, a predetermined number of trials, where each trail includes: randomly selecting the particular number of neural network models; utilizing each neural network model of the particular number of neural network models to generate a respective predictive output based on the second input data; utilizing the LR model to generate a trial output based on the respective predictive output, and determining a model trial performance based on: the trial output, the second output data, and at least one machine learning performance metric. A chain oracle model from the candidate chain oracle models is determined based on the machine learning performance metric.

Abstract: Disclosed herein is a web-based videoconference system that allows for two-dimensional screen sharing within the virtual environment. In some embodiments, data specifying a three-dimensional virtual space. The three-dimensional virtual space comprises a plurality of participants and an avatar representing each of the plurality of participants. A presentation stream is received from a first client device of a first participant. The presentation stream is mapped onto a three-dimensional model of a presentation screen in the three-dimensional virtual space. A selection of the presentation screen is received from a second participant. A two-dimensional view of the presentation stream is rendered to the second participant.

Abstract: Proposed is a federated learning system. The federated learning system comprises: a central server configured to transmit at least one global parameter of a global model to each client device, receive at least one local parameter of a local model trained from each of client devices, and update the global model using the at least one local parameter; and a plurality of client devices configured to train the local model by applying a loss between a predicted value of the global model and a predicted value of the local model possessed by itself to a loss function, and transmit at least one local parameter of the trained local model to the central server.

Abstract: A method for obfuscating deep learning (DL) models includes the step of training a DL model to obtain weights of operation (OP) layers in the trained DL model. The DL model includes an interface to a public application programming interface (API) that provides access to a compiler of an artificial intelligence (AI) processor. The method further includes the steps of obfuscating the DL model by changing a structure of the OP layers to produce an obfuscated DL model, and publishing the obfuscated DL model for access by devices. The obfuscated DL model is executable by the AI processor after compilation by the compiler on an edge device.