Abstract: Computer implemented methods for training a Student Neural Network, SNN, and for managing an environment of a communication network using a trained SNN are disclosed. The SNN is for generating an action prediction matrix for an environment in a communication network, the action prediction matrix comprising action predictions for a plurality of nodes or resources in the environment. The training method comprises using a Reinforcement Learning process to train a Teacher Neural Network, TNN, to generate an action prediction for a resource or node in the environment, and using the trained TNN to generate a first training data set including action predictions for individual nodes or resources. The training method further comprises generating a second training data set from the first training data set such that the second training data set includes action prediction matrices, and using the second training data set to update values of the parameters of the SNN.

Abstract: Some embodiments provide a method for training parameters of a network. The method receives an MT network having multiple layers of computation nodes that use first sets of weight values. Each computation node of a first layer computes an output value by computing an intermediate value based on a set of input values and a set of weight values and computing a node output value by applying an activation function to the intermediate value. The method replaces the first layer with a second layer of computation nodes followed by a third layer of computation nodes. The replacement involves (i) decomposing the sets of weight values of the first layer into second sets of weight values for the second layer and third sets of weight values for the third layer and (ii) inserting into the network a set of additional activation functions for the second layer.

Abstract: The present disclosure is directed to machine learning model architectures which provide full attention capability in each attention head while maintaining low computation and memory complexity. Specifically, according to one aspect of the present disclosure, example attention models provided herein can treat the self-attention mechanism as a conditional expectation over embeddings at each location and approximate the conditional distribution with a structured factorization. Each location can attend to all other locations, either via direct attention, or through indirect attention to group representations, which are again conditional expectations of embeddings from corresponding local regions.

Abstract: Various systems and methods are presented regarding utilizing a spiral resonator to enhance coupling between a first inductor loop and a second inductor loop to enable coupling between a first qubit and a second qubit. Operation of the first inductor loop can be controlled by a flux-tunable TCQ coupler, wherein flux-tuning can adjust operation from an OFF state (no coupling between the first qubit and the second qubit) to an ON state (the first qubit and second qubit are coupled). The spiral resonator can be located at the center of, and in the same plane as the loop of the first inductor loop. The spiral resonator can enhance inductive coupling between the first loop inductor and the second loop inductor.

Abstract: A system includes a sensor, a plurality of cells, and a processor. The sensor includes a plurality of transducers arranged on a planar surface and includes a first transducer and a second transducer. The first transducer is configured to produce an analog output signal corresponding to a detected input signal. The cells are arranged in a network and include a first and a second cell and are disposed proximate the sensor in a three-dimensional stacking fashion. The first transducer and the second transducer are electrically coupled to the first cell and the second cell in one-to-one relation. The first cell includes a plurality of inputs and a first cell output. Each input is coupled to an output of a corresponding plurality of neighboring cells. The first cell includes a first memristor and a bridge circuit configured to receive the analog output signal and provide a current corresponding to the detected input signal in a pixel-parallel fashion.

Abstract: Methods, systems, and apparatus for performing an entangling operation on a system of qubits. In one aspect, a method includes operating the system of qubits, wherein the system of qubits comprises: a plurality of first qubits, a plurality of second qubits, a plurality of qubit couplers defining nearest neighbor interactions between the first qubits and second qubits, wherein the system of qubits is arranged as a two dimensional grid and each qubit of the multiple first qubits is coupled to multiple second qubits through respective qubit couplers, and wherein operating the system of qubits comprises: pairing multiple first qubits with respective neighboring second qubits; performing an entangling operation on each paired first and second qubit in parallel, comprising detuning each second qubit in the paired first and second qubits in parallel.

Abstract: A computer-implemented method in a model-based reinforcement learning (RL) system with logic states includes switching an agent between a first mode and a second mode, the first mode being a probabilistic planning mode and the second mode being an information gathering mode. In response to the agent being in the probabilistic planning mode, the agent computes a predictive state representation, given a history of observations and actions taken, and the agent scores action candidates based on planning with the predictive state representation so that actions with resolved plans with confidence to achieve a goal state are scored higher than actions without resolved plans. In response to the agent being in the information gathering mode, the agent scores action candidates based on a Q function of a value of expected information to be gathered from a given pair of state and action.

Abstract: Methods and computational devices for the implementation thereof that enable real-time Context Resets in conversational artificial intelligence (AI) systems are provided. The methods and computational devices improve the functioning of conversational AI systems by reducing non-productive computation, preventing propagation of irrelevant Active Context, and optimizing memory use during AI sessions. These operations restore conversational coherence by reducing Misalignment or Confusion between User Inputs and System Outputs. Accordingly, the invention provides a measurable improvement to computer functionality through adaptive, User-guided context management.

Abstract: Provided is a method for classifying data in an electronic apparatus, including obtaining target data, obtaining first classification information by using a classifier set including a plurality of classifiers based on the target data, obtaining second classification information by using a neural network model based on the target data, comparing the first classification information and the second classification information, and verifying the classifier set based on a result of comparing the first classification information and the second classification information.

Abstract: A device, computer program and computer-implemented method for machine learning. The method comprises providing a task comprising an action space of a multi-armed bandit problem or a contextual bandit problem and a distribution over rewards that is conditioned on actions, providing a hyperprior, wherein the hyperprior is a distribution over the action space, determining, depending on the hyperprior, a hyperposterior for that a lower bound for an expected reward on future bandit tasks has as large a value as possible, when using priors sampled from the hyperposterior, and wherein the hyperposterior is a distribution over the action space.

Abstract: The described technology is generally directed towards automated development of machine learning pipelines. An automated framework can extract topics from a data science workspace such as a machine learning notebook, transform and annotate cells of the machine learning notebook to various machine learning pipeline stages, and orchestrate the machine learning pipeline stages in a workflow that can be deployed into production data infrastructures.

Abstract: A computer-implemented method includes receiving an incorrect prediction output by a trained machine learning model, which has been trained using training data items. The method includes identifying a training data item used to train the model that is a cause of the incorrect prediction, by determining an impact on performance of the trained machine learning model associated with removing the training data item from the plurality of training data. The trained model can then be updated to remove the effect of the identified training data item, allowing the model to be automatically corrected in view of poor quality training data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing a machine learning task on a network input to generate a network output. In one aspect, one of the systems includes an attention neural network configured to perform the machine learning task, the attention neural network including one or more attention layers, each attention layer comprising an attention sub-layer and a feed-forward sub-layer that applies an element-wise multiplication between two vectors generated as a result of two different linear transformations performed on the same attended layer input.

Abstract: Techniques described herein involve providing an automated continual learning system for machine learning models. Embodiments include generating, using a machine learning model, an output based on a sample input provided to the machine learning model. Embodiments include evaluating the output based on comparing the output to an associated training output. Embodiments include creating a natural language rule based on the evaluating using a language processing machine learning model and storing the natural language rule in a rule library. Embodiments include receiving an input to the machine learning model and retrieving, in response to the input, one or more relevant rules from the rule library. Embodiments include generating, using the machine learning model, a response based on the input and the one or more relevant rules and performing an action based on the response.

Abstract: A machine learning system and a machine learning method capable of selecting a pretrained model to be used in transfer learning in a short time without actually executing the transfer learning includes a pretrained model acquisition unit which acquires a pretrained model from a pretrained model storage unit storing a plurality of pretrained models obtained by learning a transfer source task under respective conditions; a transfer learning dataset storage unit configured to store dataset related to a transfer target task; a pretrained model adaptability evaluation unit configured to evaluate adaptability of each pretrained model acquired by the pretrained model acquisition unit to the dataset related to the transfer target task; and a transfer learning unit configured to execute, based on an evaluation result of the pretrained model adaptability evaluation unit, transfer learning using a selected pretrained model and the dataset, and outputs a learning result as a trained model.

Abstract: Techniques for training an agricultural inference machine learning model to generate valid agricultural inferences of agricultural conditions based on ground truth sensor data that falls within a plurality of ground truth sensor value ranges associated with a particular agricultural area, and to generate invalid or ambiguous agricultural inferences of agricultural conditions based on ground truth sensor data that falls outside of the plurality of ground truth sensor value ranges associated with a particular agricultural area. The agricultural inference machine learning model is trained, based on ground truth sensor data for the particular agricultural area, to determine if the subsequently received ground truth sensor data falls within or outside of that plurality of ground truth sensor value ranges that correspond to the particular agricultural area.

Abstract: An agentic workflow system and method generate question and answer pairs and prompts that may be used to aligns generative artificial intelligence (a large language model (LLM) or a large multimodal model (LMM)) with the principles of a specific domain so that the generative artificial intelligence is better able to respond to a user query in the specific domain. The system and method may also generate aligning processes that may be used to post-train an already trained generative artificial intelligence system or fine tune the training of the generative artificial intelligence system to align that generative artificial intelligence system with the principles of the specific domain. The system and method may be used to align the generative artificial intelligence system to a plurality of different domains.

Abstract: Some embodiments provide a method for improving structural sparsity of a machine-trained (MT) network. The method receives a network having multiple layers. Each layer of a set of the layers includes multiple filters of weight values. The method replaces the filters of a particular layer of the network with (i) a first set of filters of weight values, (ii) a set of scale values for the first set of filters, and (iii) a second set of filters of weight values. Each scale value corresponds to a different one of the filters of the first set of filters. The method trains the network by applying constraints to bias at least a subset of the scale values towards zero. When a particular scale value falls below a threshold value, the particular scale value is set to zero.

Abstract: Some embodiments provide a method for training a machine-trained (MT) network. The method receives a network having multiple layers. Each layer of a set of the layers includes multiple weight values. The method trains the network by alternately (1) propagating inputs through the network to generate outputs and adjusting the weight values based on differences between the generated outputs and expected outputs and (2) identifying sets of the weight values for removal according to a set of constraints that accounts for (i) a total number of weight values and (ii) an amount of time required to execute the network on a particular type of integrated circuit.

Abstract: A system and method for neural network confidence regularization is disclosed. A classification system uses a neural network training model to generate prediction. The confidence of the neural network training model is adjusted based on feature prevalence. The system processes mixed data types (binary, categorical, continuous, and date) through type-specific transformations and tensor construction. A tempering factor is calculated from the unweighted sum of features and applied to intermediate neural network outputs. This tempering mechanism reduces model confidence when several low-weight features are present, enabling faster convergence, better generalization, and improved classification accuracy compared to standard neural networks, particularly for complex non-linear relationships in tabular data domains.