Abstract: Systems and methods are provided for assisting the navigation function based in part on detection of disconnection from vehicle integration functions by other devices navigating to the same destination, where the disconnections occurred outside of a predetermined boundary associated with the destination.

Abstract: Disclosed is an operation method of a neural network including a first network and a second network, the method including acquiring state information output from the first network based on input information, determining whether the state information satisfies a condition using the second network, iteratively applying the state information to the first network in response to determining that the state information does not satisfy the condition, and outputting the state information in response to determining that the state information satisfy the condition.

Abstract: Systems, methods, and computer-readable media for achieving privacy for both data and an algorithm that operates on the data. A system can involve receiving an algorithm from an algorithm provider and receiving data from a data provider, dividing the algorithm into a first algorithm subset and a second algorithm subset and dividing the data into a first data subset and a second data subset, sending the first algorithm subset and the first data subset to the algorithm provider and sending the second algorithm subset and the second data subset to the data provider, receiving a first partial result from the algorithm provider based on the first algorithm subset and first data subset and receiving a second partial result from the data provider based on the second algorithm subset and the second data subset, and determining a combined result based on the first partial result and the second partial result.

Abstract: An autonomous module for processing stored data includes a multithreaded processor core (MPC) and a plurality of autonomous memories. Each of the plurality of autonomous memories has a memory bank, a data operator (DO) configured to implement a plurality of selectable memory behaviors, an autonomous memory operator (AMO) configured to implement a state machine to control the memory bank independently of the MPC, and at least one memory input/output (IO) port communicatively coupled with the memory bank, the AMO, and the DO. The at least one memory IO port is configured to receive a read instruction from the AMO, retrieve data from the memory bank, and send the data to the DO. The DO is configured to implement one of the plurality of selectable memory behaviors to update the data and send the updated data to the AMO via the at least one memory IO port.

Abstract: Examples described herein include systems and methods for prioritizing workloads, such as virtual machines, to enforce quality of service (“QoS”) requirements. An administrator can assign profiles to workloads, the profiles representing different QoS categories. The profiles can extend scheduling primitives that can determine how a distributed resource scheduler (“DRS”) acts on workloads during various workflows. The scheduling primitives can be used to prioritize workload placement, determine whether to migrate a workload during load balancing, and determine an action to take during host maintenance. The DRS can also use the profile to determine which resources at the host to allocate to the workload, distributing higher portions to workloads with higher QoS profiles. Further, the DRS can factor in the profiles in determining total workload demand, leading to more efficient scaling of the cluster.

Abstract: Deep neural network (DNN) models have been widely used for user-relevance content prediction. Presented herein are embodiments of a new user-relevance framework, which may be referred as Gating-Enhanced Multi-task Neural Networks (GemNN) embodiments. Neural network-based multi-task learning model embodiments herein predict user engagement with content in a coarse-to-fine manner, which gradually reduces content candidates and allows parameter sharing from upstream tasks to downstream tasks to improve the training efficiency. Also, in one or more embodiments, a gating mechanism was introduced between embedding layers and multi-layer perceptions to learn feature interactions and control the information flow fed to MLP layers. Tested embodiments demonstrated considerable improvements over prior approaches.

Abstract: Disclosed in some examples are improved address prediction and memory preloading that leverages next-delta prediction and/or far-delta prediction for scheduling using a DNN. Previous memory access sequence data that identify one or more memory addresses previously accessed by one or more processors of a system may be processed and then converted into a sequence of delta values. The sequence of delta values are then mapped to one or more classes that are then input to a DNN. The DNN then outputs a predicted future class identifier sequence that represents addresses that the DNN predicts will be accessed by the processor in the future. The predicted future class identifier sequence is then converted back to a predicted delta value sequence and back into a set of one or more predicted addresses.

Abstract: Disclosed according to one exemplary embodiment includes not limited to: a filtering unit configured to generate filtering information by filtering a residual image corresponding to a difference between an original image and a prediction image; an inverse filtering unit configured to generate inverse filtering information by inversely filtering the filtering information; an estimator configured to generate the prediction image based on the original image and reconstruction information; a CNN-based in-loop filter configured to receive the inverse filtering information and the prediction image and to output the reconstruction information; and an encoder configured to perform encoding based on the filtering information and information of the prediction image, and wherein the CNN-based in-loop filter is trained for each of the plurality of artefact sections according to an artefact value or for each of the plurality of quantization parameter sections according to a quantization parameter.

Abstract: Various embodiments relate to data science and data analysis, computer software and systems, and computing architectures and data models configured to facilitate management of enterprise functions, and, more specifically, to an enterprise computing and data processing platform configured to activate risk management transformations of enterprise data in-situ, responsive to identifying a risk event, and further configured to implement a risk management data channel to facilitate analyses and responses associated with an enterprise computing device. In some examples, a method may include receiving a risk data signal, identifying a portion of the risk data signal, computing data representing a risk level, classifying data associated with a hierarchical business data object in accordance with a risk level, aggregating classified data with other data associated with other business data objects classified as a function of risk to form aggregated data, causing presentation of aggregated data as a function of risk.

Abstract: A method and an apparatus for exclusive reinforcement learning are provided, comprising: collecting information of states of an environment through the communication interface and performing a statistical analysis on the states using the collected information; determining a first state value of a first state among the states in a training phase and a second state value of a second state among the states in an inference phase based on analysis results of the statistical analysis; performing reinforcement learning by using one reinforcement learning unit of a plurality of reinforcement learning unit which performs reinforcement learnings from different perspectives according to the first state value; and selecting one of actions determined by the plurality of reinforcement learning unit based on the second state value and applying selected action to the environment.

Abstract: A computer-implemented method, computer program product and computing system for executing a description model when utilizing a website; detecting a failure associated with the execution of the description model; re-executing the description model one or more times in an attempt to utilize the website; and if a failure is detected one or more times, reporting the failure to a user.

Abstract: Disclosed herein are techniques related to glucose estimation without continuous glucose monitoring. In some embodiments, the techniques may involve receiving input data associated with a user. The input data may comprise discrete blood glucose measurement data associated with the user, activity data associated with the user, contextual data associated with the user, or a combination thereof. The techniques may also involve using an estimation model and the input data associated with the user to generate one or more estimated blood glucose values associated with the user.

Abstract: Embodiments of the present invention are directed to facilitating data preprocessing for machine learning. In accordance with aspects of the present disclosure, a training set of data is accessed. A preprocessing query specifying a set of preprocessing parameter values that indicate a manner in which to preprocess the training set of data is received. Based on the preprocessing query, a preprocessing operation is performed to preprocess the training set of data in accordance with the set of preprocessing parameter values to obtain a set of preprocessed data. The set of preprocessed data can be provided for presentation as a preview. Based on an acceptance of the set of preprocessed data, the set of preprocessed data is used to train a machine learning model that can be subsequently used to predict data.

Abstract: Disclosed is an electronic apparatus. The electronic apparatus may include a memory configured to store one or more training data generation models and an artificial intelligence model, and a processor configured to generate personal training data that reflects a characteristic of a user using the one or more training data generation models, train the artificial intelligence model using the personal learning data as training data, and store the trained artificial intelligence model in the memory.

Abstract: According to an embodiment, an information processing device is configured to assign a first computing device one or more first tasks of processing respective one or more first partial data of a plurality of partial data included in an n-dimensional target data, n being an integer greater than or equal to 2, the target data being to be processed using a neural network, the one or more first partial data including first data and second data adjacent to the first data in a direction of m-dimension, m being an integer satisfying 1?m?n; and instruct the first computing device to execute a second task included in the one or more first tasks, according to an execution status of second partial data of the plurality of partial data included in the target data, the second partial data being executed by the second computing device.

Abstract: Provided is a device for ensembling data received from prediction devices and a method of operating the same. The device includes a data manager, a learner, and a predictor. The data manager receives first and second device prediction results from first and second prediction devices, respectively. The learner may adjust a weight group of a prediction model for generating first and second item weights, first and second device weights, based on the first and second device prediction results. The first and second item weights depend on first and second item values, respectively, of the first and second device prediction results. The first device weight corresponds to the first prediction device, and the second device weight corresponds to the second prediction device. The predictor generates an ensemble result of the first and second device prediction results, based on the first and second item weights and the first and second device weights.

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: A highway detection system may include a telematics device associated with a vehicle having one or more sensors arranged therein, a mobile device associated with a user traveling in the vehicle, and a server computer. The server computer may receive traveling data for a trip of the user from the one or more sensors and via the telematics device. The server computer may then determine whether the user is traveling within a city or on a highway based on analysis of the traveling data for the trip of the user, which may include a statistical analysis to calculate standard deviations of metrics in the traveling data. In response to the determination, the server computer may generate a notification to transmit to the user based on whether the user is traveling within a city or on a highway and transmit the notification to the mobile device associated with the user.

Abstract: Aspects of the disclosure provide for controlling an autonomous vehicle. For instance, a first probability distribution may be generated for the vehicle at a first future point in time using a generative model for predicting expected behaviors of objects and a set of characteristics for the vehicle at an initial time expected to be perceived by an observer. Planning system software of the vehicle may be used to generate a trajectory for the vehicle to follow. A second probability distribution may be generated for a second future point in time using the generative model based on the trajectory and a set of characteristics for the vehicle at the first future point expected to be perceived by the observer. A surprise assessment may be generated by comparing the first probability distribution to the second probability distribution. The vehicle may be controlled based on the surprise assessment.

Abstract: A system, method, and computer-readable medium for performing a data center management and monitoring operation. The data center management and monitoring operation includes: receiving data center data from a plurality of data center assets within a data center, the data center data comprising outlier data detection data; assigning the data center data to a vectorized input space; reducing a dimension of the vectorized input space to a latent space; decoding the latent space to provide a vectorized decoded output space; and, performing an outlier data detection operation using the vectorized decoded output space, wherein the outlier data detection operation is performed to detect data center asset outlier data.

Abstract: Implementations relate to crop yield prediction at the field- and pixel-level. In various implementations, a first temporal sequence of high-elevation digital images may be obtained that capture a first geographic area and are acquired over a first predetermined time interval while the first geographic area includes a particular crop. A first plurality of other data points may also be obtained that influence a ground truth crop yield of the first geographic area after the first predetermined time interval. The first plurality of other data points may be grouped into temporal chunks corresponding temporally with respective images of the first temporal sequence. The first temporal sequence and the temporal chunks of the first plurality of other data points may be applied, e.g., iteratively, as input across a machine learning model to estimate a crop yield of the first geographic area at the end of the first predetermined time interval.

Abstract: A device may receive project management data associated with development of a software product and may process a first portion of the project management data, with first models, to generate timeliness scores and an overall timeliness score for the software product. The device may process a second portion of the project management data, with second models, to generate quality scores and an overall quality score for the software product and may process a third portion of the project management data, with third models, to generate product readiness scores and an overall product readiness score for the software product. The device may utilize a fourth machine learning model, with the overall timeliness score, the overall quality score, and the overall product readiness score, to generate a success probability for the software product and may perform one or more actions based on the success probability for the software product.

Abstract: Provided is a simulation method performed by a process simulator, implemented with a recurrent neural network (RNN) including a plurality of process emulation cells, which are arranged in time series and configured to train and predict, based on a final target profile, a profile of each process step included in a semiconductor manufacturing process. The simulation method includes: receiving, at a first process emulation cell, a previous output profile provided at a previous process step, a target profile and process condition information of a current process step; and generating, at the first process emulation cell, a current output profile corresponding to the current process step, based on the target profile, the process condition information, and prior knowledge information, the prior knowledge information defining a time series causal relationship between the previous process step and the current process step.

Abstract: Apparatuses and methods can be related to implementing age-based network training. An artificial neural network (ANN) can be trained by introducing errors into the ANN. The errors and the quantity of errors introduced into the ANN can be based on age-based characteristics of the memory device.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for monitoring technology usage and performance. In some implementations, use of a technology item by one or more individuals assigned to use the technology item is monitored. Based on the monitoring, usage data that indicates usage of the technology item is generated. One or more criteria for evaluating the usage of the technology item by the one or more individuals is identified. It is determined whether usage data satisfies the one or more criteria. A system provides, for display on a user interface, output data indicating whether the usage data satisfies the one or more criteria.

Abstract: In various embodiments, an unsupervised training application executes a neural network on a first point cloud to generate keys and values. The unsupervised training application generates output vectors based on a first query set, the keys, and the values and then computes spatial features based on the output vectors. The unsupervised training application computes quantized context features based on the output vectors and a first set of codes representing a first set of 3D geometry blocks. The unsupervised training application modifies the first neural network based on a likelihood of reconstructing the first point cloud, the quantized context features, and the spatial features to generate an updated neural network. A trained machine learning model includes the updated neural network, a second query set, and a second set of codes representing a second set of 3D geometry blocks and maps a point cloud to a representation of 3D geometry instances.

Abstract: The embodiments recite systems and methods that improve the traditional underwriting process within a financial institution. These embodiments produce an underwriting model that emulates the resolution patterns of top performing underwriters. The underwriting model once is built and tested is incorporated into decision tools that provide underwriters with insightful advices when underwriting a client. The embodiments use statistical learning techniques such as support vector machine and logistic regression. These techniques can assume a linear or nonlinear relationship between factors and risk classes. Furthermore, the underwriting model also uses artificial intelligence tools such as expert systems and fuzzy logic. A company's underwriting standards and best underwriting practices may be updated periodically so that underwriting model based on decision heuristic keep improving the quality of its output over time.

Abstract: An artificial neural network that includes first subnetworks to implement known functions and second subnetworks to implement unknown functions is trained. The first subnetworks are trained separately and in parallel on corresponding known training datasets to determine first parameter values that define the first subnetworks. The first subnetworks are executing on a plurality of processing elements in a processing system. Input values from a network training data set are provided to the artificial neural network including the trained first subnetworks. Error values are generated by comparing output values produced by the artificial neural network to labeled output values of the network training data set. The second subnetworks are trained by back propagating the error values to modify second parameter values that define the second subnetworks without modifying the first parameter values. The first and second parameter values are stored in a storage component.

Abstract: A system may forecast a plurality of workload measurements for a database management system (DBMS) at respective times based on a workload model. The system may determine, based on the forecasted workload measurements, configuration parameter sets optimized for the DBMS at the respective times. The system may generate a reconfiguration plan. The system may determine performance that would result from reconfiguring nodes of the DBMS with the configurations parameter sets. The system may select a reconfiguration plan in response to the performance satisfying a fitness criterion. The system may cause, at the reconfiguration times, the nodes to begin reconfiguration with the configuration parameter sets included in the selected reconfiguration plan.

Abstract: The presently disclosed subject matter includes an apparatus with a processor and a memory storing code which, when executed by the processor, causes the processor to receive a data profile associated with a candidate resource, the data profile includes a set of attributes of the candidate resource which are relevant for assessing the candidate resource's suitability to satisfy a particular resource demand. The apparatus extracts an n-dimensional feature vector from the received data profile, the n-dimensional feature vector capturing aspects of the candidate resource's attributes and process said n-dimensional feature vector with a first ensemble machine learning model to generate a first suitability factor. Likewise, the apparatus process said n-dimensional feature vector with a second ensemble machine learning model to generate a second suitability factor. The apparatus determines whether to allocate the candidate resource to the particular resource demand using said first and second suitability factors.

Abstract: A compile method for a neural network, the compile method includes receiving data related to the neural network, generating a grouped layer by grouping layers comprised in the neural network based on the data, generating a set of passes executable in parallel based on a dependency between a plurality of passes to process the neural network, generating a set of threads performing a plurality of optimization functions based on whether optimization operations performed by the optimization functions is performed independently for the layers, respectively, or sequentially based on a dependency between the layers, and performing compilation in parallel based on the grouped layer, the set of passes, and the set of threads.

Abstract: The disclosed computer-implemented method may include (i) receiving a first transport request and a second transport request, (ii) evaluating a fitness of matching the first and second transport requests to be fulfilled by a transport provider, based at least partly on a transportation overlap between the first and second transport requests, (iii) generating a simulated future transport request, (iv) evaluating a fitness of matching the first transport request with the simulated future transport request, based at least in part on a transportation overlap between the first transport request and the simulated future transport request, and (v) matching the first and second transport requests based at least in part on the fitness of matching the first and second transport requests and based at least in part on the fitness of matching the first transport request with the simulated future transport request. Various other methods, systems, and computer-readable media are disclosed.

Abstract: Aspects of the disclosure relate to using machine learning for resource planning. A computing platform may detect an occupancy modification event for a physical space. Based on detecting the occupancy modification event, the computing platform may send commands directing display of a data collection prompt to end user devices, which may prompt for work to be performed by users of the end user devices in the physical space during a first day. Using natural language processing, the computing platform may analyze user input information and other occupancy data to determine whether or not the users of the end user devices have permission to occupy the physical space during the first day. The computing platform may cause the end user devices to display a resource management interface indicating whether or not the users of the end user devices have valid permission to physically occupy the physical space during the first day.

Abstract: Tiered advertisement bidding is disclosed. One or more quality metrics associated with a user profile are determined. An advertisement bid is selected from a plurality of tiered bids based at least in part on the determined quality metrics. Determining the quality metrics can include determining a conversion assessment. Determining the quality metric can also include determining whether a need associated with a user profile has been met for a category. In some cases, persona detection is performed with respect to the user profile.

Abstract: Some embodiments provide a method for training a machine-trained (MT) network that processes inputs using network parameters. The method propagates a set of input training items through the MT network to generate a set of output values. The set of input training items comprises multiple training items for each of multiple categories. The method identifies multiple training item groupings in the set of input training items. Each grouping includes at least two training items in a first category and at least one training item in a second category. The method calculates a value of a loss function as a summation of individual loss functions for each of the identified training item groupings. The individual loss function for each particular training item grouping is based on the output values for the training items of the grouping. The method trains the network parameters using the calculated loss function value.

Abstract: Disclosed herein are system, method, and computer program product embodiments for the detection of human presence in an energy efficient manner using a plurality of sensors such as those of a battery-powered device such as a television remote, and a device with a processor, such as a television. Data gathered from an initial television WiFi radio scan, or an initial low-powered detection scan from the television remote, may be analyzed by the processor to determine a potential presence of one or more humans are present proximate to the device. If there is such a potential presence, the device remote can enter a full-powered detection mode to accurately determine the presence or absence of one or more humans, and take further actions.

Abstract: In selected embodiments a recommendation generator builds a network of interrelationships between venues, reviewers and users based on attributes and reviewer and user reviews of the venues. Each interrelationship or link may be positive or negative and may accumulate with other links (or anti-links) to provide nodal links the strength of which are based on commonality of attributes among the linked nodes and/or common preferences that one node, such as a reviewer, expresses for other nodes, such as venues. The links may be first order (based on a direct relationship between, for instance, a reviewer and a venue) or higher order (based on, for instance, the fact that two venue are both liked by a given reviewer). The recommendation engine in certain embodiments determines recommended venues based on user attributes and venue preferences by aggregating the link matrices and determining the venues which are most strongly coupled to the user.

Abstract: Provided are methods for efficient and optimal feature extraction, which can include identifying observations that satisfy a predetermined configuration of metrics. Some methods described also include extracting feature values of a respective metric from the identified observations and determining parameters for feature extraction based on the feature values extracted from the identified observations and a respective label of the identified observations. Systems and computer program products are also provided.

Abstract: Techniques that facilitate feedback loop learning between artificial intelligence systems are provided. In one example, a system includes a monitoring component and a machine learning component. The monitoring component identifies a data pattern associated with data for an artificial intelligence system. The machine learning component compares the data pattern to historical data patterns for the artificial intelligence system to facilitate modification of at least a component of the artificial intelligence system and/or one or more dependent systems of the artificial intelligence system.

Abstract: A method includes receiving historical interaction data, which includes a plurality of historical interactions. Each historical interaction is associated with a plurality of data fields. The method includes assigning a plurality of weights to the plurality of data fields, generating a neural network using the plurality of weights and the plurality of data fields, identifying a first plurality of feature indicators indicative of a first class, the first class being different from a second class; receiving a second plurality of feature indicators derived from data relating to compromised accounts, updating, a probability distribution component using the first plurality of feature indicators and the second plurality of feature indicators, and receiving current data for an interaction. The method also includes applying the probability distribution component to the current data, and scoring the interaction using the probability distribution component.

Abstract: In one embodiment, a method includes receiving an image associated with an object in an environment, the image being captured by sensors associated with a vehicle, generating a feature representation of the image, determining a potential ground control point associated with the object based on the feature representation of the image, determining a predetermined location reading based on the potential ground control point, calculating a differential relative to the predetermined location reading based on the potential ground control point, and determining a location of the vehicle based on the differential and the predetermined location reading based on the potential ground control point.

Abstract: The present disclosure relates to the field of laboratory diagnostics. Specifically, methods are disclosed for determining a patient's risk of suffering from heart failure (HF) based on the detection of NT-proBNP, troponin T, and/or a natriuretic peptide. Also disclosed are methods for improving both the accuracy and speed of HF risk models by incorporating biomarker data from patient samples.

Abstract: Machine learning systems for predictive targeting and optimizing engagement are described herein. In various embodiments, the system includes 1) training a first machine learning computer model to generate machine predicted outcomes; (2) determining weights based on the machine predicted outcomes; (3) generating a second machine learning computer model based on the weights; and (4) generating machine learned predictions for candidates.

Abstract: A computer-implemented method, system, and non-transitory machine readable medium for a graph-based analysis for an Information Technology (IT) operations includes generating a temporal graph by extracting one or more of operation objects, relations and attributes from operation data of workloads distributed across a plurality of levels of the IT operation within a predetermined time window. Anomalies are detected from the extracted operation data and annotating corresponding objects in the graph. A directional impact between corresponding objects on the temporal graph is determined, and the temporal graph is refined based on the determined directional impact. Accessible paths in the temporal graph indicating error propagation are searched, and potential causes for the detected anomalies in the temporal graph are identified. A list of the potential causes of the anomalies is generated, and a root cause ranked for each of the corresponding objects in the temporal graph.

Abstract: A computer-implemented method is provided for data augmentation. The method includes calculating, by a hardware processor for each of words in a text data, a word replacement probability based on a word occurrence frequency in the text data, wherein the word replacement probability decreases with increasing word occurrence frequency. The method additionally includes selectively replacing at least one of the words in the text data with words predicted therefor by a Bidirectional Neural Network Language Model (BiNNLM) to generate augmented text data, based on the word replacement probability.

Abstract: The present disclosure relates to systems and methods for optimizing benefits plan options offered by an organization through balancing derived population preferences with organizational preferences by analyzing historical selections made by individuals. Census data dividing members of an organization into census divisions may be applied to machine learning algorithm(s) to derive estimated selection preferences of the members. Using selection preferences, costs of various product offering scenarios and overall member satisfaction estimates of the scenarios may be calculated. Product offering scenarios meeting member preference criteria and organizational budget criteria may be presented for review.

Abstract: Disclosed according to one exemplary embodiment includes not limited to: a filtering unit configured to generate filtering information by filtering a residual image corresponding to a difference between an original image and a prediction image; an inverse filtering unit configured to generate inverse filtering information by inversely filtering the filtering information; an estimator configured to generate the prediction image based on the original image and reconstruction information; a CNN-based in-loop filter configured to receive the inverse filtering information and the prediction image and to output the reconstruction information; and an encoder configured to perform encoding based on the filtering information and information of the prediction image, and wherein the CNN-based in-loop filter is trained for each of the plurality of artefact sections according to an artefact value or for each of the plurality of quantization parameter sections according to a quantization parameter.

Abstract: Artificial-intelligence-based river information system. In an embodiment, a first training dataset is used to train a travel time prediction model to predict a travel time along the waterway for a given trip. In addition, a second training dataset is used to train a river level prediction model to predict a river level along the waterway for a given time. For each of a plurality of trips, a request is received that specifies the trip and a time of the trip, and, in response to the request, the travel time prediction model is used to predict a travel time for the trip, and the river level prediction model is used to predict a river level of the waterway at one or more points along the trip. Then, a voyage plan is generated based on one or both of the predicted travel time and the predicted river level.

Abstract: Aspects described herein provide a method including: receiving input data at a machine learning model, comprising: a plurality of processing layers; a plurality of gate logics; a plurality of gates; and a fully connected layer; determining based on a plurality of gate parameters associated with the plurality of gate logics, a subset of the plurality of processing layers with which to process the input data; processing the input data with the subset of the plurality of processing layers and the fully connected layer to generate an inference; determining a prediction loss based on the inference and a training label associated with the input data; determining an energy loss based on the subset of the plurality of processing layers used to process the input data; and optimizing the machine learning model based on: the prediction loss; the energy loss; and a prior probability associated with the training label.

Abstract: A machine learning system includes a coach machine learning system that uses machine learning to help a student machine learning system learn its system. By monitoring the student learning system, the coach machine learning system can learn (through machine learning techniques) “hyperparameters” for the student learning system that control the machine learning process for the student learning system. The machine learning coach could also determine structural modifications for the student learning system architecture. The learning coach can also control data flow to the student learning system.