Abstract: Implementations for training a denoising stacked autoencoder (DAE) using a noisy training dataset comprising a noisy sub-set and a non-noisy sub-set, providing an artificial neural network (ANN) including multiple hidden layers, at least one hidden layer including at least a portion of an encoder of the DAE, the at least a portion of the encoder comprising parameters determined during training of the DAE, training the ANN using a training dataset, and providing a version of the ANN for inference.

Abstract: Systems, methods, and devices generate predictive data segments using machine learning and natural language processing. The method receives training data with digital identifiers and binary classification labels indicating target outcomes. It accesses external data sources containing textual attribute descriptions, which are preprocessed by concatenating multiple descriptions per identifier for large language model input. The model outputs high-dimensional vector embeddings encoding semantic relationships. A neural network trains using these embeddings and classification labels to predict probability scores. For new data, the method preprocesses textual descriptions, generates embeddings via the language model, and applies the neural network to produce probability scores representing conversion propensity values. It selects digital identifiers with scores exceeding a configurable threshold and transmits them to a digital platform for prioritized resource allocation.

Abstract: A computer-implemented method for estimating uncertainties using a neural network, in particular, a neural process, in a model. The model models a technical system and/or a system behavior of the technical system. An architecture of the neural network for estimating uncertainties is also described.

Abstract: A computer-implemented method, a computer program product, and a computer system for accelerated learning. A computer partitions an independent variable of input data into partitions. A computer creates data samples in each of the partitions, where each of the data samples has less granularity. For each of the partitions, a computer trains a machine learning model independently on each of the data samples and compares results of training on the data samples. For each of one or more partitions in which results of training on a predetermined number of the data samples are statistically identical at a predetermined confidence level, a computer outputs a result of training on one of the data samples. For each of one or more rest partitions in which no result of training has been outputted, a computer merges each pair of the data samples and uses merged data samples to train the machine learning model.

Abstract: A 3D compute-in-memory accelerator system and method for efficient inference of Mixture of Expert (MoE) neural network models. The system includes a plurality of compute-in-memory cores, each in-memory core including multiple tiers of in-memory compute cells. One or more tiers of in-memory compute cells correspond to an expert sub-model of the MoE model. One or more expert sub-models are selected for activation propagation based on a function-based routing, the tiers of the corresponding experts being activated based on this function. In one embodiment, this function is a hash-based tier selection function used for dynamic routing of inputs and output activations. In embodiments, the function is applied to select a single expert or multiple experts with input data-based or with layer-activation-based MoEs for single tier activation. Further, the system is configured as a multi-model system with single expert model selection or with a multi-model system with multi-expert selection.

Abstract: A method performed by a first node in a distributed network is provided. The method includes receiving a request from a second node for transfer learning of a machine learning model for a use case. The request includes a description of a local environment and a use case identifier at the second node. The method further includes identifying a matching criteria for the local environment of the second node based on the use case. The method further includes determining whether at least one distributed node from a plurality of distributed nodes in the distributed network satisfies at least one of a match or a closest match to the matching criteria.

Abstract: The present disclosure provides a method including receiving a target object for detection and a predetermined area in which the target object is to be detected. The predetermined area is associated with a plurality of content capturing devices. The target object is detected within one or more frames of video data. Responsive to detecting the target object, detection data is determined, including: a particular content capturing device associated with the one or more frames of video data, a location of the particular content capturing device, a time at which the one or more frames were captured, and/or weather data associated with the geolocation of the particular content capture device at the time. The present detection data is provided to an Artificial Intelligence model to predict a next geolocation at which the target object is likely to be detected and/or a timeframe for detection of the target object.

Abstract: A computer-implemented method including: receiving a trained machine learning model; extracting a set of features associated with the machine learning model, wherein each of the extracted features is assigned a feature importance score which represents a relative explanatory power of the feature with respect to an output of the machine learning model; generating a set of marginal queries based, at least in part, on a selected subset of the features having a highest the feature importance score; performing a measurement of the set of marginal queries on a source database, to obtain measurements of the set of marginal queries on the source database; and using the measurements to generate synthetic data that matches the measurements.

Abstract: A system and methods for multivariant learning and optimization repeatedly generate self-organized experimental units (SOEUs) based on the one or more assumptions for a randomized multivariate comparison of process decisions to be provided to users of a system. The SOEUs are injected into the system to generate quantified inferences about the process decisions. Responsive to injecting the SOEUs, at least one confidence interval is identified within the quantified inferences, and the SOEUs are iteratively modified based on the at least one confidence interval to identify at least one causal interaction of the process decisions within the system. The causal interaction can be used for testing, diagnosis, and optimization of the system performance.

Abstract: An apparatus and method for generating a sequence output. The apparatus includes at least a processor and a memory communicatively connected to the at least a processor. The memory instructs the processor to receive one or more target datums of target data associated with an entity and a plurality of entity data, classify the plurality of entity data based on a category of data, weigh each classified datum by assigning a weight to classified data as a function of predicted return data and the target data, encode the weighted data of the plurality of entity data into a plurality of actions wherein each action corresponds to one or more target datums, and generate a sequence output as a function of a temporal datum of a plurality of temporal data associated with each of the one or more target datums of the target data and the plurality of actions.

Abstract: An electronic device may execute a neural network model for generating an image. Generating the image includes consecutively obtaining a plurality of frames by using a camera. A first frame among the plurality of frames may be presented to a first group of the neural network model, as a first input, and first result data corresponding to the first input stored in a memory. The first frame may be presented to a second group, as a second input. Upon determining that a second computation parameter of the second group is the same as a first computation parameter of the first group, the first result data is used as second result data corresponding to the second input, without performing a neural network computation that is based on the second group.

Abstract: Continuous condition monitoring of an electrical system, and in particular for early fault detection, is provided. The condition monitoring unit is formed with an interface to a memory in which a trained normal state model is stored as a one-class model that has been trained in a training phase with normal state data and represents a normal state of the electrical system. Further, the condition monitoring unit comprises a data interface for continuously acquiring sensor data of the electrical system by means of a set of sensors, an extractor for extracting features from the acquired sensor data, a differentiator for determining deviations of the extracted features from learned features of the normal state model by means of a distance metric (e.g., Euclidean norm, sum norm, maximum norm), a scoring for calculating an anomaly score from the determined deviations, and an output unit for outputting the calculated anomaly score.

Abstract: The present disclosure provides a method for target identification and scoring, the method comprising: detecting an object within one or more frames of a content stream; deriving one or more images of the object from the content stream; normalizing the one or more images; processing the normalized one or more images to determine a species of the object; identifying an Artificial Intelligence (“AI”) module corresponding to the species of the object; virtually regenerating the object based on the following: the species of the object, the normalized one or more images, and one or more of the following: physical orientation of the object, time of object detection, and illumination of the object; providing the regenerated object to the AI module configured to perform object recognition; receiving an identification of the object from the AI module; and updating an object profile with object identification data.

Abstract: Systems, methods, and other embodiments associated with a data quality framework for a machine learning pipeline are described. In one embodiment, a method includes receiving an input dataset prior to training a machine learning model with the input dataset. The input dataset, having data records in column tabular form, is analyzed to determine a data quality of the input dataset prior to training the machine learning model. A data quality score is generated that represents an overall quality of the input dataset, where the data quality score is generated based on a combined ensemble of at least two factors selected from (i) an abundance factor, (ii) a completeness factor, and (iii) a dimension efficiency factor. Based on the data quality score, the input dataset is either permitted or prohibited from continuing in the machine learning pipeline, and corrective actions may be performed.

Abstract: A schedule creation method is a method for creating a time schedule by executing a learning step multiple times. The learning step includes sequentially placing patterns each indicating a procedure in a processing sequence in a timetable for defining a time schedule for respective elements of a substrate processing apparatus. The sequentially placing patterns in a timetable includes: acquiring one or more placeable patterns that are allowed to be placed in the timetable from among the patterns based on a prescribed constraint condition; predicting and selecting through machine learning a pattern that makes an evaluation value maximum from among the one or more placeable patterns; and updating the timetable by placing the selected pattern in the timetable.

Abstract: A learning model optimization device includes a binarization matrix setting unit configured to set a binarization matrix m in which each element is a numerical value of “0” or “1” and a transformed matrix setting unit configured to set a transformed matrix M having, as an element, a product of each element of the parameter matrix and each element of the binarization matrix in the same row and the same column. The learning model optimization device further includes a learning unit configured to perform machine learning using the transformed matrix M and change a numerical value of each element of the binarization matrix m such that a result of the machine learning approaches teacher data, thereby optimizing the binarization matrix m, and a re-randomization processing unit configured to change again a parameter of the parameter matrix w.

Abstract: In a method of creating a learning model using a controller configured to perform pruning on a neural network, the pruning includes a first pruning process in which a pruning process is performed in units of channels of convolutional layers and a second pruning process in which a pruning process is performed in units of weight parameters.

Abstract: Apparatuses, systems, and techniques to determine whether to remove one or more neural network layers. In at least one embodiment, one or more neural network layers are determined to be removed based on, for example, a neural architecture search (NAS).

Abstract: A method of slicing a deep learning model for a heterogeneous embedded system includes collecting, by a model slicing apparatus, an execution time and power consumption when each layer corresponding to one layer of a deep learning model including a plurality of layers is executed in each computing device of the heterogeneous embedded system, predicting, by the model slicing apparatus, a performance cost and a power cost when each of the layers is executed in each of the computing devices using the execution time and the power consumption, predicting, by the model slicing apparatus, a communication cost when transmitting information from each of the layers to a next layer in each of the computing devices, and slicing, by the model slicing apparatus, the plurality of layers so that different sliced layers are allocated to each of the computing devices based on the performance cost, the power cost, and the communication cost in a given execution time limit condition using a reinforcement learning model.

Abstract: Systems, methods, and computer-readable medium are provided for healthcare analysis. Data corresponding to a plurality of patients is received. The data is parsed to generate normalized data for a plurality of variables, with normalized data generated for more than one variable for each patient. A causal relationship network model is generated relating the plurality of variables based on the generated normalized data using a Bayesian network algorithm. The causal relationship network model includes variables related to a plurality of medical conditions or medical drugs. In another aspect, a selection of a medical condition or drug is received. A sub-network is determined from a causal relationship network model. The sub-network includes one or more variables associated with the selected medical condition or drug. One or more predictors for the selected medical condition or drug are identified.