Abstract: In at least one aspect, a method comprises: receiving information associated with a plurality of scheduled actions, each of the plurality of scheduled actions defining a device to be controlled, an action to be carried out with respect to the device to be controlled, a scheduled day on which to carry out the scheduled action and a scheduled time of day to carry out the scheduled action on the scheduled day; determining a chronological ordering that indicates a chronological order in which at least a subset of the plurality of scheduled actions are scheduled to be carried out; generating a view based at least in part on the chronological ordering; and displaying the generated view. Such methods may be performed or implemented by various devices, apparatuses and/or systems.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting an action to be performed by a reinforcement learning agent interacting with an environment. A current observation characterizing a current state of the environment is received. For each action in a set of multiple actions that can be performed by the agent to interact with the environment, a probability distribution is determined over possible Q returns for the action-current observation pair. For each action, a measure of central tendency of the possible Q returns with respect to the probability distributions for the action-current observation pair is determined. An action to be performed by the agent in response to the current observation is selected using the measures of central tendency.

Abstract: Described are techniques for determining a data policy suitable for association with a data object based on the data access pattern for the data object. Correspondence between the data access pattern of the data object and pattern data, indicative of data access patterns stored in association with data policies, may be determined. Based on the correspondence between the data access pattern of the data object and a particular data access pattern of the pattern data, the data policy associated with the particular data access pattern may be suitable for use with the data object. A set of suitable data policies may be refined based on the content or metadata associated with the data object and the code or deployment status of services that access the data object. Once the access pattern for a data object is known, subsequent interactions with the data object may be analyzed to identify anomalous traffic.

Abstract: A computer-implemented method, system, and computer program product are provided for activity recognition in a mobile device. The method includes receiving a plurality of unlabeled videos from one or more cameras. The method also includes generating a classified video for each of the plurality of unlabeled videos by classifying an activity in each of the plurality of unlabeled videos. The method additionally includes storing the classified video in a location in a memory designated for videos of the activity in each of the classified videos.

Abstract: An identifying information assignment system includes: a generating portion configured to generate, for a learning result obtained by attaining a predetermined capability through a predetermined learning process by machine learning, identifying information for identifying the predetermined learning process; and an assignment portion configured to assign the generated identifying information to the learning result.

Abstract: A computer-implemented method, system, and computer program product are provided for activity recognition in a surveillance system. The method includes receiving a plurality of unlabeled videos from one or more cameras. The method also includes classifying an activity in each of the plurality of unlabeled videos. The method additionally includes controlling an operation of a processor-based machine to react in accordance with the activity.

Abstract: An apparatus may be configured to obtain, for a Siamese neural network having a recurrent neural network (RNN), an initial representation associated with a target object at a first time step and a set of candidate regions at a current time step. The apparatus may determine an updated representation associated with the target object based on the initial representation at the first time step and observed information associated with the target object at a set of previous time steps, and the observed information associated with the target object may be represented by a hidden state of the RNN. The apparatus may output the updated representation associated with the target object for matching with the set of candidate regions at the current time step by the Siamese neural network. The apparatus may determine the updated representation further based on a hidden state at a previous time step.

Abstract: The present invention discloses methods, systems, and tools for knowledge processing and visualizations by building various data structures corresponding to uncovered informational data such as values of association strengths or significance measures and maps of ontological subjects of compositions or one or more content accompanying a request for service by a user. In one embodiment of the invention the method assigns and calculates an ontological subject association strength/value measures and spectrums to each composition or ontological subjects of the composition. The resulting data, spectrums, and the adjacency matrix of the map or visualization are used to evaluate the merits of the compositions in the context of reference universes. It is also used as a research guiding tool for knowledge discovery or automatically generating high value compositions or new or less known knowledge about the ontological subjects of the universe.

Abstract: A messaging includes a sender and a receiver of text messages via a central unit that includes a database, in which a user dataset containing a univocal identification code for each user relates with an element dataset containing univocal identification codes of all graphic and/or audio elements manageable by the messaging system, thereby defining the graphic and/or audio elements accessed or managed by each user. Each graphic and/or audio element is manageable by association with the univocal identification code. A first software module in the sender or receiver sends a data structure to the central unit based on a graphic and/or audio element and has a first field containing a univocal identification code, and a second software module in the central unit creates a structured data package based on the received data structure having a second field filled at least partly with the first field of the data structure.

Abstract: A method, system, and apparatus configured to identify discriminating features in a plurality of applications, determine via code analysis, when a first application is subjected to classification, positions of the first application's code that correspond to the discriminating features, and forward to a classification algorithm, such that according to its output the code fragments corresponding to the discriminating features are reported.

Abstract: Synthetic training information/data of a second probe style is generated based on first probe information/data of a first probe style using a style transfer model. First probe information/data is defined. An instance of first probe information/data comprises labels and first probe style sensor information/data. A style transfer model generates training information/data based on at least a portion of the first probe information/data. An instance of training information/data corresponds to an instance of first probe information/data and comprises second probe style sensor information/data. The first and second probe styles are different. A second probe style model is trained using machine learning and the training information/data. The second probe style model is used to analyze second probe style second probe information/data to extract map information/data from the second probe information/data. Each instance of second probe data is captured by one or more second probe sensors of a second probe apparatus.

Abstract: In some examples, a system for training a neural network can include a processor to detect a trained neural network application. The processor can also detect a set of images, wherein the neural network application is not trained with the set of images. Additionally, the processor can train an adapter network based on the trained neural network application and the set of images, wherein the adapter network is to be trained by freezing weights of the trained neural network and modifying weights of the adapter network. Furthermore, the processor can use the trained adapter network to process at least one additional image, the processed additional image to be transmitted to the trained neural network to generate an output signal.

Abstract: A method, computer program product and/or system is disclosed. According to an aspect of this invention, one or more processors receive a query of a first database, where the query includes: (i) an operand, and (ii) an operator indicating a distance-based similarity measure. One or more processors further determine a result set based on the query, wherein the result set includes a plurality of records, and wherein a record is included in the result set based on a vector nearest-neighbor computation between: (i) a first vector corresponding to the operand, and (ii) a second vector corresponding to the record, wherein the second vector is included in a vector space model that is based on a textual representation of the first database.

Abstract: Embodiments for continuous time alignment of a collection of independent sensors monitoring a common entity by one or more processors. One or more activity events associated with a monitored entity may be identified in the time series sensor data collected from a plurality of sensors. The one or more activity events may be dynamically characterized in the time series sensor data using a machine learning operation. The time series data streams from each of the plurality of sensors may be time-aligned by aligning the one or more activity events.

Abstract: The present disclosure is directed to training, and providing recommendations via, a temporal ensemble of neural networks. The neural networks in the temporal ensemble can be trained at different times. For example, a neural network can be periodically trained using current item interaction data, for example once per day using purchase histories of users of an electronic commerce system. The item interaction data can be split into a more recent group and a less recent group, for example the last two weeks of data and the remainder of the last two years of data. The periodic training of neural networks, using updated data and the sliding windows created by the date split, results in a number of different models for predicting item interaction events. Using a collection of these neural networks together as a temporal ensemble can increase recommendation accuracy without requiring additional hardware for training.

Abstract: Systems and methods for determining an over power subscription adjustment for a Radio Equipment (RE) are provided. In some embodiments, a method of operation of an RE includes determining an over power subscription adjustment for one or more carriers based on one or more system parameters and providing the over power subscription adjustment to a Radio Equipment Controller (REC) for the one or more carriers. According to some embodiments, this permits the network operator to better utilize radio network resources by over subscribing carrier power levels. The operator can do this without exposing radio hardware to damage by adaptive learning of Mean Power Limiting (MPL) parameters. Further, the degree of over subscription can be dynamically adjusted to suit network operating conditions. This can lead to improved quality of service metrics and increased radio sector cell size.

Abstract: System, methods, and other embodiments described herein relate to conditionally generating custom images by sampling latent space of a generator. In one embodiment, a method includes, in response to receiving a request to generate a custom image, generating a component instruction by translating a description about requested characteristics for the object instance into a vector that identifies a portion of a latent space within a respective generator. The method includes computing the object instance by controlling the respective one of the generators according to the component instruction to produce the object instance. The respective one of the generators being configured to generate objects within a semantic object class. The method includes generating the custom image from at least the object instance to produce the custom image from the description as a photorealistic image approximating a real image corresponding to the description.

Abstract: Systems and methods for use in training a convolutional neural network (CNN) for image and video transformations. The CNN is trained by adding noise to training data set images, transforming both the noisy image and the source image, and then determining the difference between the transformed noisy image and the transformed source image. The CNN is further trained by using an object classifier network and noting the node activation levels within that classifier network when transformed images (from the CNN) are classified. By iteratively adjusting the CNN to minimize a combined loss function that includes the differences between the node activation levels for the transformed references images and when transformed source are classified and the differences between the transformed noisy image and the transformed source image, the artistic style being transferred is maintained in the transformed images.

Abstract: In one embodiment, a computing system is configured to track objects in an environment or localize a user device. For example, the system accesses an image of an environment captured from a viewpoint. Based on the image, the system detects landmarks that are associated with objects in the environment and identifies expected landmarks that are expected to be observable from the viewpoint using a landmark database. The system determines that at least one of the expected landmarks is currently unobservable in the environment by comparing the expected landmarks with the detected landmarks. By accessing semantic information associated with the at least one expected landmark, the system updates the landmark database based on a determination that the semantic information satisfies predetermined criteria and removes the at least one expected landmark from the landmark database. The system performs object tracking, object mapping, or re-localization within the environment using the updated landmark database.

Abstract: A system comprising at least one hardware processor; and a non-transitory computer-readable storage medium having stored thereon program instructions, the program instructions executable by the at least one hardware processor to: receive, as input, a plurality of electronic documents, apply a trained machine learning classifier to automatically classify at least some of said plurality of electronic documents, wherein said machine learning classifier comprises two or more attention layers, and wherein at least one of the attention layers comprises an adjustable parameter which controls a distribution of attention weights assigned by said attention layer.

Abstract: Techniques are provided herein for creating well-balanced computer-based reasoning systems and using those to control systems. The techniques include receiving a request to determine whether to use one or more particular data elements, features, cases, etc. in a computer-based reasoning model (e.g., as data elements, cases or features are being added, or as part of pruning existing features or cases). Conviction measures (such as targeted or untargeted conviction, contribution, surprisal, etc.) are determined and inclusivity conditions are tested. The result of comparing the conviction measure can be used to determine whether to include or exclude the feature, case, etc. in the computer-based reasoning model. A controllable system may then be controlled using the computer-based reasoning model.

Abstract: A system is provided for natural language processing. In some embodiments, the system includes an encoder for generating context-specific word vectors for at least one input sequence of words. The encoder is pre-trained using training data for performing a first natural language processing task. A neural network performs a second natural language processing task on the at least one input sequence of words using the context-specific word vectors. The first natural language process task is different from the second natural language processing task and the neural network is separately trained from the encoder. In some embodiments, the first natural processing task can be machine translation, and the second natural processing task can be one of sentiment analysis, question classification, entailment classification, and question answering.

Abstract: Techniques are provided herein for creating well-balanced computer-based reasoning systems and using those to control systems. The techniques include receiving a request to determine whether to use one or more particular features, cases, etc. in a computer-based reasoning model (e.g., as cases or features are being added, or as part of pruning existing features or cases). Conviction measures (such as targeted or untargeted conviction, contribution, surprisal, etc.) are determined and inclusivity conditions are tested. The result of comparing the conviction measure can be used to determine whether to include or exclude the feature, case, etc. in the computer-based reasoning model. A controllable system may then be controlled using the computer-based reasoning model. Examples controllable systems include self-driving cars, image labeling systems, manufacturing and assembly controls, federated systems, smart voice controls, automated control of experiments, energy transfer systems, and the like.

Abstract: Techniques are provided herein for creating well-balanced computer-based reasoning systems and using those to control systems. The techniques include receiving a request to determine whether to use one or more particular features, cases, etc. in a computer-based reasoning model (e.g., as cases or features are being added, or as part of pruning existing features or cases). Conviction measures (such as targeted or untargeted conviction, contribution, surprisal, etc.) are determined and inclusivity conditions are tested. The result of comparing the conviction measure can be used to determine whether to include or exclude the feature, case, etc. in the computer-based reasoning model. A controllable system may then be controlled using the computer-based reasoning model. Examples controllable systems include self-driving cars, image labeling systems, manufacturing and assembly controls, federated systems, smart voice controls, automated control of experiments, energy transfer systems, and the like.

Abstract: The invention utilizes a two-component system to detect third party security threats and drive internal system processes based on the detection. The first component of the system is a threat level engine, which collects external and internal system data on a real-time basis to determine changes in conditions that may give rise to a security threat. Based on the external and internal data, the level engine may calculate a threat assessment level to determine the level of the threat. The second component of the system is a third party analytics engine, which may comprise a machine learning component which is configured to detect threat patterns and anomalies, which may in turn be used to trigger events or to drive internal system processes.

Abstract: An information processing apparatus includes a storing unit that stores a neural network that recognizes a data characteristic of data, and a correction unit that corrects the data by using the neural network stored in the storing unit such that the neural network recognizes that the data characteristic of the data is a predetermined data characteristic.

Abstract: Methods, systems, and computer program products are provided for processing a request regarding relationships among instances of entities. A graphical representation of instances of entities is generated and includes one or more source nodes, each representing an instance of an input entity of a request, and one or more related nodes, each representing an instance of a second entity related to one or more corresponding instances of the input entity and associated with a corresponding confidence score for the relationship. Each of the one or more related nodes associated with a confidence score satisfying a threshold is identified. One or more supplemental nodes are added to the graphical representation, each of which represents a corresponding instance of a third entity with a relationship to a corresponding instance of the second entity. The graphical representation is traversed to identify relationships between instances of entities and produce results for the request.

Abstract: Systems, methods, and computer readable media related to training and/or utilizing a neural network model to determine, based on a sequence of sources that each have an electronic interaction with a given electronic resource, one or more subsequent source(s) for interaction with the given electronic resource. For example, source representations of those sources can be sequentially applied (in an order that conforms to the sequence) as input to a trained recurrent neural network model, and output generated over the trained recurrent neural network model based on the applied input. The generated output can indicate, for each of a plurality of additional sources, a probability that the additional source will subsequently (e.g., next) interact with the given electronic resource. Such probabilities indicated by the output can be utilized in performance of further electronic action(s) related to the given electronic resource.

Abstract: A reconfigurable neural network circuit is provided. The reconfigurable neural network circuit comprises an electronic synapse array including multiple synapses interconnecting a plurality of digital electronic neurons. Each neuron comprises an integrator that integrates input spikes and generates a signal when the integrated inputs exceed a threshold. The circuit further comprises a control module for reconfiguring the synapse array. The control module comprises a global final state machine that controls timing for operation of the circuit, and a priority encoder that allows spiking neurons to sequentially access the synapse array.

Abstract: An object detection device extracts feature for input data utilizing an encoder, the input data including labeled data and unlabeled data and detects object in each of the input data, utilizing an object detector. The object detection device generates region data for each of the input data, each of the region data corresponding to the detected object and generates restoration data from the region data and meta-information related to the detected object for each of the input data utilizing a decoder corresponding to the encoder. The object detection device executes learning of the encoder and the object detector based on a result detected by the object detector and a label associated with the input data, when the input data is labeled data. The object detection device executes learning of the encoder, the object detector, and the decoder, based on the input data and the restoration data.

Abstract: Given a number of records and a number of target classes to which these records belong to, a (weakly) supervised machine learning classification method leverages known possibly dirty classification rules, efficiently and accurately learns a classification model from training data, and applies the learned model to the data records to predict their classes.

Abstract: A control method includes: executing a first process that includes obtaining movement information on movement of an object from an image sequence; executing a second process that includes presuming a set of a candidate region in which an object may be present in a first image in the image sequence and an index indicating probability of presence of the object in the candidate region according to a feature amount of the first image; and executing a third process that includes determining a position of the object in the first image by using the set of the candidate region and the index presumed, wherein the second process includes correcting information obtained during the presuming based on the obtained movement information.

Abstract: The present disclosure relates to a method and attention neural network for automatically learning embeddings for various latent aspects of textual claims and documents performed in an attention neural network comprising one or more latent aspect models for guiding an attention mechanism of the neural network, wherein the method comprises the steps of inserting a claim document pair, in each of the latent aspect models and a latent aspect vector to select significant sentences to form document representations for each respective latent aspect of the latent aspect vector, concatenating the document representations to establish an overall document representation, calculating a class probability distribution by means of the overall document representation, and classifying the claim of document as true or false using the class probability distribution.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a neural network system used to control an agent interacting with an environment to perform a specified task. One of the methods includes causing the agent to perform a task episode in which the agent attempts to perform the specified task; for each of one or more particular time steps in the sequence: generating a modified reward for the particular time step from (i) the actual reward at the time step and (ii) value predictions at one or more time steps that are more than a threshold number of time steps after the particular time step in the sequence; and training, through reinforcement learning, the neural network system using at least the modified rewards for the particular time steps.

Abstract: The technology disclosed herein enables the generation of malware traffic signatures by performing natural language processing on known malware traffic using a neural network. In a particular embodiment, a method provides generating sentences comprising first information obtained from a plurality of fields in each of a plurality of known malware data packets in a first malware family. The method further provides inputting the sentences into a first neural network for natural language processing of the sentences and generating one or more signatures for the first malware family from results of the natural language processing of the sentences.

Abstract: Processing implemented by computer means of sound data output by at least one sensor and activation of at least one actuator of an acoustically radiating structure. The sensor senses an acoustic signal output by the vibration of the radiating structure. The radiating structure bears at least one actuator controlled by the computer means and is thus involved in the vibration of the radiating structure. In particular, the method comprises the steps of: a) measuring a transfer function of the actuator, radiating structure and sensor assembly, b) controlling activation of the actuator so as to make the radiating structure vibrate, according to a selected setpoint: taking account of the transfer function measured, and taking account of the acoustic signal sensed by the sensor in feedback mode.

Abstract: Exemplary embodiments may use word embeddings to enhance scanning of programming code scripts for sensitive subject matter, such as confidential subject matter. The scanning may be performed by a neural network in some exemplary embodiments. The neural network initially may be trained on a corpus of programming code scripts to identify keywords relating to sensitive subject matter, such as passwords, tokens or credentials. The neural network may not only identify instances of the keywords but also may identify related terms as well. The output of the scan may be a ranked list of terms in the programming code script that may relate to sensitive subject matter.

Abstract: Methods, systems, and computer program products for systematic browsing of automated conversation exchange program knowledge bases are provided herein. A computer-implemented method includes generating one or more initial questions to be submitted to an automated conversation exchange program; generating one or more natural language variants of the initial questions; submitting the one or more natural language variants of the initial questions to the automated conversation exchange program; identifying one or more valid responses returned by the automated conversation exchange program in response to the submitted natural language variants of the initial questions; deriving one or more items of data from the identified valid responses; storing the derived items of data in an automated conversation exchange program knowledge base; and systematically browsing the automated conversation exchange program knowledge base in connection with one or more application tasks.

Abstract: A recognition system includes: a sensor processing unit (SPU) that performs sensing to output a sensor value; a task-specific unit (TSU) including an object detection part that performs an object detection task based on the sensor value and a semantic segmentation part that performs a semantic segmentation task based on the sensor value; and a generic-feature extraction part (GEU) including a generic neural network disposed between the sensor processing unit and the task-specific unit, the generic neural network being configured to receive the sensor value as an input to extract a generic feature to be input in common into the object detection part and the semantic segmentation part.

Abstract: In an example, a transitive local predicates may be inferred across multiple joins, wherein the multiple outer joins include multiple outer joins. A query connecting tables by the multiple joins is analyzed. A local predicate may then be identified from the analyzed query and may be determined to be either a where-clause local predicate or an on-clause local predicate. Accordingly, a transitive local predicate may be inferred across a selection of the tables based on the determination.

Abstract: A computer-implemented method for domain analysis comprises: obtaining, by a computing device, a domain; and inputting, by the computing device, the obtained domain to a trained detection model to determine if the obtained domain was generated by one or more domain generation algorithms. The detection model comprises a neural network model, a n-gram-based machine learning model, and an ensemble layer. Inputting the obtained domain to the detection model comprises inputting the obtained domain to each of the neural network model and the n-gram-based machine learning model. The neural network model and the n-gram-based machine learning model both output to the ensemble layer. The ensemble layer outputs a probability that the obtained domain was generated by the domain generation algorithms.

Abstract: A passive infra-red automatic fever detection system and method for identifying individuals presenting a thermal profile indicative of an abnormally elevated body temperature, either after an initial thermal reading or after the initial thermal reading and one or more subsequent thermal readings. The system and method may be configured for automatic detection, notification, and alarming of individuals with elevated body temperatures that may be indicative of an infection.

Abstract: A medical scan image analysis system is operable to receive a plurality of medical scans that represent a three-dimensional anatomical region and include a plurality of cross-sectional image slices. A plurality of three-dimensional subregions corresponding to each of the plurality of medical scans are generated by selecting a proper subset of the plurality of cross-sectional image slices from each medical scan, and by further selecting a two-dimensional subregion from each proper subset of cross-sectional image slices. A learning algorithm is performed on the plurality of three-dimensional subregions to generate a fully convolutional neural network. Inference data corresponding to a new medical scan received via the network is generated by performing an inference algorithm on the new medical scan by utilizing the fully convolutional neural network. An inferred abnormality is identified in the new medical scan based on the inference data.

Abstract: A method for processing tensor data for pattern recognition and a computer device are provided. The method includes: constructing a decision function by the optimal projection tensor W which has been rank-one decomposed together with the offset scalar b, and inputting to-be-predicted tensor data which has been rank-one decomposed into the decision function for prediction.

Abstract: Aspects are generally directed to network security systems and methods of monitoring network activity. In one example, a network security system includes and interface to receive a Hypertext Transfer Protocol (HTTP) network log that includes a matrix of data, a feature extraction component configured to extract a connectivity matrix from the HTTP network log based on a recurring pattern within the matrix of data, and a training module configured to provide deep learning architecture training data based on the connectivity matrix. The system may include a deep learning architecture configured to receive and propagate the training data through one or more layers thereof to train the one or more layers, and being configured to generate a general data representation of the HTTP network log. The system may include a behavior analytics component to detect a discordant network activity within the HTTP network log based on the general data representation.

Abstract: An asset class type of a new asset is predicted or determined based upon an evaluation of time series data from the new asset. A predicted asset type is used to identify sensors of the new asset to use for data collection. Using the readings of selected sensors from the new asset, states of the new asset are obtained. The duration at least one of these states of the new asset is determined. This information can be subsequently used to optimize the performance of the new asset.

Abstract: A system and method for generating recommendations for improving online advertising success of an image-based advertisement are provided. The method includes identifying at least one visual characteristic of the advertisement; classifying the advertisement into at least one advertisement category based on the identified at least one visual characteristic; analyzing a plurality of advertisements belonging to the at least one advertising category to identify at least one visual characteristic associated with successful advertisements; generating at least one recommendation for improving the image-based advertisement based on the identified at least one successful advertisement visual characteristic.

Abstract: A system for reducing analog noise in a noisy channel, comprising: an interface configured to receive analog channel output comprising a stream of noisy binary codewords of a linear code; and a computation component configured to perform the following: for each analog segment of the analog channel output of block length: calculating an absolute value representation and a sign representation of a respective analog segment, calculating a multiplication of a binary representation of the sign representation with a parity matrix of the linear code, inputting the absolute value representation and the outcome of the multiplication into a neural network for acquiring a neural network output, and estimating a binary codeword by component-wise multiplication of the neural network output and the sign representation.

Abstract: A system and method for batched, supervised, in-situ machine learning classifier retraining for malware identification and model heterogeneity.

Abstract: Systems and methods for estimating a height of an object from a monocular image are described herein. Objects are detected in the image, each object being indicated by a region of interest. The image is then cropped for each region of interest and the cropped image scaled to a predetermined size. The cropped and scaled image is then input into a convolutional neural network (CNN), the output of which is an estimated height for the object. The height may be represented by a mean of a probability distribution of possible sizes, a standard deviation, as well as a level of confidence. A location of the object may be determined based on the estimated height and region of interest. A ground truth dataset may be generated for training the CNN by simultaneously capturing a LIDAR sequence with a monocular image sequence.