Abstract: There is provided a system and method of automated design space determination for deep neural networks. The method includes obtaining a teacher model and one or more constraints associated with an application and/or target device or process used in the application configured to utilize a deep neural network; learning an optimal architecture using the teacher model, constraints, a training data set, and a validation data set; and deploying the optimal architecture on the target device or process for use in the application.

Abstract: The present disclosure relates to an improved machine learning-based recommender system and method for cold-start predictions on an ecommerce platform. The improved system predicts user-item interactions with respect to cold-start items in which only side information is available. Item representations generated by an item neural network encoder from item side information are shared with a user neural network. The item representations are used, along with user feedback history, to generate user representations. Specifically, a weight matrix in the first layer of the user neural network encoder is fixed with the shared item embeddings. The effect of this is that, when the user neural network encoder is applied to an input user-item interaction vector, the output of the first layer of the user neural network encoder is a function of the item representations of the items for which the user provided positive feedback.

Abstract: Multilingual neural machine translation systems having monolingual adapter layers and bilingual adapter layers for zero-shot translation include an encoder configured for encoding an input sentence in a source language into an encoder representation and a decoder configured for processing output of the encoder adapter layer to generate a decoder representation. The encoder includes an encoder adapter selector for selecting, from a plurality of encoder adapter layers, an encoder adapter layer for the source language to process the encoder representation. The decoder includes a decoder adapter selector for selecting, from a plurality of decoder adapter layers, a decoder adapter layer for a target language for generating a translated sentence of the input sentence in the target language from the decoder representation.

Abstract: Disclosed is a Bayesian classification recognition system based on an industrial PaaS platform, comprising: an IaaS infrastructure service layer, a G-PaaS graph neural network processing layer, an O-PaaS docking service layer and an SaaS system application layer. The G-PaaS graph neural network processing layer is configured for point cloud feature generation, point cloud feature learning, point cloud structure estimation and point cloud model classification; and the recognition accuracy of a workpiece point cloud model is improved through the Bayesian classification recognition system based on the industrial PaaS platform.

Abstract: Systems, devices, and methods are provided for training and/or inferencing using machine-learning models. In at least one embodiment, an identity embedding and a content embedding are extracted from training media data, such as training audio data and/or training visual data. The identity embedding may encode information relating to a person's voice and/or visual properties. A decoder may be trained to create synthetic media, such as audio and visual content, based on identity and content embeddings.

Abstract: Certain aspects of the present disclosure provide techniques for parameterized activation functions. Input data is processed with at least one layer of the neural network model comprising a parameterized activation function, and at least one trainable parameter of the parameterized activation function is updated based at least in part on output from the at least one layer of the neural network model. The at least one trainable parameter may adjust at least one of a range over which the parameterized activation function is nonlinear or a shape of the parameterized activation function, and/or may adjust a location of at least one pivot of the parameterized activation function.

Abstract: A system and method for configuring a device includes using a machine learning model to generate a user behavior model based on user behavior data. The user behavior data may include time series data collected from user interactions with a first device, and the machine learning model may include a classification model configured to classify the user behavior data into the one or more classifications. A mapping may be created by training a machine learning model, using user behavior models from a plurality of users and device settings from the plurality of users, to identify one or more relationships between device settings and classifications of the user behavior data. The system and method configures one or more settings of a second device based on the user behavior model and the mapping.

Abstract: A method and system for determining an access score is disclosed. The method includes receiving an access request to access a resource by a user device. Next, a user embedding is retrieved from an embedding table, the user embedding associated with a user identifier of the user device and providing a multidimensional data point that represents a context of a user identifier. The context may correspond to the user identifier appearing in previous access requests within temporal proximity to other access requests from a subset of other user devices among a plurality of user devices. The method then inputs the user embedding into a first machine learning model that is trained based at least in part on the embedding table. The first machine learning model subsequently outputs an access score that corresponds to a level of authenticity of authorizing the user device to access the resource.

Abstract: Systems and methods for automated generation and customization of interfaces are disclosed herein. The method includes receiving a request to launch an interface and identifying a subject of the interface and a viewing user of the interface. Based on attributes of one or both of the subject and the viewing user, a page is selected and customized. One or several containers associated with the page are identified. Data associated with the containers is retrieved and customized, and the container is likewise customized. A customized view is generated based on the customized page, the customized one or more containers, and the customized data.

Abstract: At a first terminal of a structure capable of hosting Majorana Zero Modes, a first set of data points measuring conductance between the first terminal and a middle terminal of the structure is obtained for different values of bias voltage at the first terminal and at least one other parameter. At a second terminal of the structure, a second set of data points measuring conductance between the second terminal and the middle terminal is obtained for different values of bias voltage at the second terminal and of the at least one other parameter. A measure of mutual information is obtained between the first and second data sets. It is determined whether a signature consistent with a pair of Majorana Zero Modes is present in the structure based on the measure of mutual information. The method may be carried out by a quantum computer.

Abstract: A method for applying machine learning to an application includes: a) generating a candidate policy by a learner; b) executing a program in at least one simulated application based on a set of candidate parameters provided based on the candidate policy and a state of the at least one simulated application, execution of the program providing interim results of tested sets of candidate parameters based on a measured performance information of the execution of the program; c) collecting a predetermined number of interim results and providing an end result based on a combination of the candidate parameters and/or the state with the measured performances information by a trainer; and d) generating a new candidate policy by the learner based on the end result.

Abstract: The present invention provides a hardware-friendly framework for implementing a point-of-care diagnosis hardware tool for practical end-user convenience, power saving and resource utilization. The hardware tool is non-invasive and comfortable for the patient, as a primary means of differential diagnosis between two neuromuscular diseases such as neuropathy and myopathy. The provided hard-ware tool comprises a feature extractor configured to receive electrodiagnostic signals (preferably EMG signals) of a patient and extract one or more features from the collected signals; and a classifier configured to receive the extracted features and classify a neuromuscular disease for the patient based on the extracted features. The classifier is a single layer machine-learning perceptron trained with datasets consisted of electrodiagnostic signals of patients to perform a linearly separable binary classification.

Abstract: A method for performing contrastive learning for graph tasks and datasets by employing an information-aware graph contrastive learning framework is presented. The method includes obtaining two semantically similar views of a graph coupled with a label for training by employing a view augmentation component, feeding the two semantically similar views into respective encoder networks to extract latent representations preserving both structure and attribute information in the two views, optimizing a contrastive loss based on a contrastive mode by maximizing feature consistency between the latent representations, training a neural network with the optimized contrastive loss, and predicting a new graph label or a new node label in the graph.

Abstract: A method for performing contrastive learning for graph tasks and datasets by employing an information-aware graph contrastive learning framework is presented. The method includes obtaining two semantically similar views of a graph coupled with a label for training by employing a view augmentation component, feeding the two semantically similar views into respective encoder networks to extract latent representations preserving both structure and attribute information in the two views, optimizing a contrastive loss based on a contrastive mode by maximizing feature consistency between the latent representations, training a neural network with the optimized contrastive loss, and predicting a new graph label or a new node label in the graph.

Abstract: A method for performing contrastive learning for graph tasks and datasets by employing an information-aware graph contrastive learning framework is presented. The method includes obtaining two semantically similar views of a graph coupled with a label for training by employing a view augmentation component, feeding the two semantically similar views into respective encoder networks to extract latent representations preserving both structure and attribute information in the two views, optimizing a contrastive loss based on a contrastive mode by maximizing feature consistency between the latent representations, training a neural network with the optimized contrastive loss, and predicting a new graph label or a new node label in the graph.

Abstract: The disclosure describes embodiments of systems, methods, and non-transitory computer readable storage media that dynamically surface multiple user-specific application function GUIs for setting up (or accessing) application functions for a user account. For instance, the disclosed systems can select and display a GUI for a first application function from a set of application functions. Moreover, upon receiving user interactions with the GUI for the first application function (and identifying whether or not the first application system is activated for the user account), the disclosed systems can select a second application function from the set of application functions and displays a GUI (e.g., a setup interface) for the second application function from the set of application functions.

Abstract: A method for performing contrastive learning for graph tasks and datasets by employing an information-aware graph contrastive learning framework is presented. The method includes obtaining two semantically similar views of a graph coupled with a label for training by employing a view augmentation component, feeding the two semantically similar views into respective encoder networks to extract latent representations preserving both structure and attribute information in the two views, optimizing a contrastive loss based on a contrastive mode by maximizing feature consistency between the latent representations, training a neural network with the optimized contrastive loss, and predicting a new graph label or a new node label in the graph.

Abstract: A system, method, and a computer program product for generating custom models are provided. A custom model platform includes a user interface that receives selections for model criteria, including a platform model group, a risk profile, a sequence of transformations, and a sequence of constraints. A platform model is retrieved based on the platform model group and the risk profile. The platform model is transformed into a custom model using the sequence of transformations. The custom model is constrained using the sequence of constraints. The custom model and the platform model are displayed on the user interface.

Abstract: The disclosed technique incudes a method for dynamically selecting an algorithm that predicts usage of a mobile application based on network signals. In one example, an application profile includes known network signals and designates a best predictive algorithm. When network signals of user devices are subsequently captured, the best algorithms of sufficiently matching profiles are used to estimate application usage. As such, for example, the popularity of a particular application or relationships among applications can be determined for managing the network or for commercial purposes.

Abstract: An information processing method according to the present disclosure includes the steps of: evaluating, by a computer, a neural network having a structure held in a divided manner by a first device and a second device based on information on transfer of information between the first device and the second device in the neural network; and determining, by the computer, the structure of the neural network based on the evaluation of the neural network.