Abstract: A system for context-aware decision making of an autonomous agent includes a computing system having a context selector and a map. A method for context-aware decision making of an autonomous agent includes receiving a set of inputs, determining a context associated with an autonomous agent based on the set of inputs, and optionally any or all of: labeling a map; selecting a learning module (context-specific learning module) based on the context; defining an action space based on the learning module; selecting an action from the action space; planning a trajectory based on the action S260; and/or any other suitable processes.

Abstract: Techniques are disclosed to migrate data via query co-evaluation. In various embodiments, an input data associated with a source database S and a target schema T to which the input data is to be migrated is received. A set of relational conjunctive queries from target schema T to source database S is received. Query co-evaluation is performed on the received set of relational conjunctive queries to transition data from source database S to target schema T.

Abstract: Described herein are methods and systems for obtaining high-quality images, which may be suitable for various applications such as training and operating artificial intelligence (AI) systems. Specifically, a system may comprise multiple cameras and one or more actuators that are capable of moving these cameras relative to each other. For example, these cameras may form one or more stereo pairs, each pair having its stereo axis. The actuators can change baselines in these pairs and/or tilt these stereo axes relative to the imaged objects to address possible self-similarity issues associated with the shape of these objects and their orientation relative to the cameras. In some examples, the simultaneous images captured by these cameras are used to construct a three-dimensional (3D) model. The fidelity of this model is then used to determine the position of the cameras (as a camera unit or individually for each camera).

Abstract: A method comprises constructing, for a specific activity, a decision model based on action data of one or more actions, each action having an identifier of an activity performed with a computer application, use data being operated on by the computer application, and a timestamp, the decision model including a rule specifying a first activity as a next activity or a probabilistic classifier that accepts use data being operated on via the specific activity and outputs an identifier of an activity to be performed as a next activity with a probability. The method comprises detecting performance of the specific activity, when the decision model for the specific activity includes the rule, automatically applying the rule, and when the decision model for the specific activity includes the probabilistic classifier and no applicable rule: executing the probabilistic classifier to obtain a list of candidate next activities and an associated list of probabilities.

Abstract: The present disclosure relates to methods, systems, and storage media for detecting events based on updates to node profiles from electronic activities. Exemplary implementations may access an electronic activity transmitted or received via an electronic account associated with a data source provider; generate a plurality of activity field-value pairs; maintain a plurality of node profiles; identify a first state of a first node profile of the plurality of node profiles; update the first node profile using the electronic activity; identify a second state of the first node profile subsequent to updating the first node profile using the electronic activity; detect a state change of the first node profile based on the first state and the second state; determine that the state change satisfies an event condition; and store an association between the first node profile and an event type corresponding to the event condition.

Abstract: A speech image providing method according to an embodiment includes generating a standby state image in which a person is in a standby state, generating a plurality of back-motion images at a preset frame interval from the standby state image for image interpolation between a preset reference frame of the standby state image, generating a speech state image in which a person is in a speech state based on a source of speech content, returning the standby state image being played to the reference frame based on the plurality of back-motion images of the standby state image, based on a point of time when the generating of the speech state image is completed, and generating a synthetic speech image in combination with frames of the speech state image from the reference frame.

Abstract: A visual guidance system for a vehicle includes an imaging system for producing a digital image of an environment, a three-dimensional scanning system for producing a digital point cloud of the environment, and a memory storing instructions executable by a processor to: process the digital image to detect an object and classify the object; process the point cloud to group points into a grouping representing the object; fuse the point cloud grouping with the digital image object to produce a fused data set frame, and determine the object location from the fused data set frame; determine an object velocity relative to the vehicle by comparing the fused data set frame to a previous data set frame; determine whether the object is an obstacle based on the object location and relative velocity; determine a threat level of the obstacle; and report a threat if the threat level exceeds a threshold.

Abstract: Techniques are disclosed for extracting micro-features at a pixel-level based on characteristics of one or more images. Importantly, the extraction is unsupervised, i.e., performed independent of any training data that defines particular objects, allowing a behavior-recognition system to forgo a training phase and for object classification to proceed without being constrained by specific object definitions. A micro-feature extractor that does not require training data is adaptive and self-trains while performing the extraction. The extracted micro-features are represented as a micro-feature vector that may be input to a micro-classifier which groups objects into object type clusters based on the micro-feature vectors.

Abstract: A package includes a first die with a compute element and first region, a second die with a compute element and second region, and a bridging element connecting the first and second dies. The bridging element includes interconnect regions for electrical coupling, a first photonic path from the first interconnect region to the second, and a second photonic path in the reverse direction. A photonic transceiver is integrated into the bridging element, with one portion sending and receiving optical signals via the photonic paths, and the other portion located in an AMS block in the first die or second die near the memory and compute elements. The transceiver portions are connected by a short electrical interconnect (less than 2 mm).

Abstract: A neural network model replaces the supervised labeling component of a supervised learning system with an incremental cluster validity index-based unsupervised labeling component. An implementation is presented combining fuzzy adaptive resonance theory predictive mapping (ARTMAP) and incremental cluster validity indices (iCVI) for unsupervised machine learning purposes, namely the iCVI-ARTMAP. An iCVI module replaces the adaptive resonance theory (ART) module B of a fuzzy ARTMAP neural network model and provides assignments of input samples to clusters (i.e., labels) at each learning iteration in accordance to any of several possible iCVI methods described. A map field incrementally builds a many-to-one mapping of the categories of ART module A to the cluster labels. At the end of each learning epoch, clusters may be merged and/or split using the iCVI, which is recomputed incrementally except for the newly cluster during a split.

Abstract: Provided is a process including: receiving a data token to be passed from a first node to a second node; retrieving machine learning model attributes from a collection of one or more of the sub-models of a federated machine-learning model; determining based on the machine learning model attributes, that the data token is learning relevant to members of the collection of one or more of the sub-models and, in response, adding the data toke to a training set to be used by at least some members of the collection of one or more of the sub-models; determining a collection of data tokens to transmit from the second node to a third node of the set of nodes participating in a federated machine-learning model; and transmitting the collection of data tokens.

Abstract: Described herein are systems, methods, and non-transitory computer readable media for using 3D point cloud data such as that captured by a LiDAR as ground truth data for training an instance segmentation deep learning model. 3D point cloud data captured by a LiDAR can be projected on a 2D image captured by a camera and provided as input to a 2D instance segmentation model. 2D sparse instance segmentation masks may be generated from the 2D image with the projected 3D data points. These 2D sparse masks can be used to propagate loss during training of the model. Generation and use of the 2D image data with the projected 3D data points as well as the 2D sparse instance segmentation masks for training the instance segmentation model obviates the need to generate and use actual instance segmentation data for training, thereby providing an improved technique for training an instance segmentation model.

Abstract: A method for verifying a worker agent includes receiving, by a core node, from a worker agent, a capability description describing a plurality of tasks and, for each of the plurality of tasks, (i) at least one parameter of the task and (ii) an outcome expected to be produced by performing the task. The method includes generating, based on the capability description, a plurality of request-output pairs, each representing a particular request and a corresponding baseline output expected to be produced upon processing the request. The core node receives, from the worker agent, a plurality of outputs, each of the plurality of outputs generated by the worker agent and corresponding to one of the plurality of request-output pairs. The core node compares the plurality of baseline outputs to the plurality of actual outputs to produce comparison output and determines whether to approve the worker agent based on the comparison output.

Abstract: An apparatus for generating a lip sync image according to disclosed embodiment has one or more processors and a memory which stores one or more programs executed by the one or more processors. The apparatus includes a first artificial neural network model configured to generate an utterance match synthesis image by using a person background image and an utterance match audio signal corresponding to the person background image as an input, and generate an utterance mismatch synthesis image by using the person background image and an utterance mismatch audio signal not corresponding to the person background image as an input, and a second artificial neural network model configured to output classification values for an input pair in which an image and a voice match and an input pair in which an image and a voice do not match by using the input pairs as an input.

Abstract: An image synthesis device according to a disclosed embodiment has one or more processors and a memory which stores one or more programs executed by the one or more processors. The image synthesis device includes a first artificial neural network provided to learn each of a first task of using a damaged image as an input to output a restored image and a second task of using an original image as an input to output a reconstructed image, and a second artificial neural network connected to an output layer of the first artificial neural network, and trained to use the reconstructed image output from the first artificial neural network as an input and improve the image quality of the reconstructed image.

Abstract: Methods and systems are provided for detecting and processing gestures, expressions (e.g., facial), tone and/or gestures of the user for the purpose of improving the quality and speed of interactions with computer-based systems. Such information may be detected by one or more sensors such as, for example, electromyography (EMG) sensors used to monitor and record electrical activity produced by muscles that are activated. Other sensor types may be used, such as optical, inertial measurement unit (IMU), or other types of bio-sensors. The system may use one or more sensors to detect speech alone or in combination with gestures, expressions (e.g., facial), tone and/or gestures of the user to provide input or control of the system.

Abstract: A system and method for inconsistency detection. A method includes semantically analyzing a first set of data to extract features. The features include subjects represented in the first set of data. Semantically analyzing the first set of data includes applying a machine learning model. The first set of data is consolidated into a knowledge base based on the extracted features. The knowledge base includes a graph having nodes and edges. The nodes represent the subjects, and the edges represent connections among the subjects. The knowledge base is queried based on a second set of data in order to obtain knowledge base query results. Querying the knowledge base includes semantically analyzing the second set of data in order to identify more subjects. Semantically analyzing the second set of data includes applying the machine learning model. Data among the second set of data is validated based on the knowledge base query results.

Abstract: A system, including an analysis server computer, a robot that is configured to participate in interactions with one or more participants; wherein the robot is configured to collect information related to the participants during the interaction, wherein the robot accepts a summary report related to the interaction from a participant or forms a summary report, wherein the robot is configured to forward the summary report to the analysis server for analysis, wherein the analysis server is configured to analyze the summary report and determine if the summary report should pass or fail, when the summary report passes the analysis server is configured to store the summary report in a database; and when the summary report fails the analysis server is configured to provide feedback to the robot with a list of problems for correction in the summary report.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for estimating wave properties of a body of water. A computer-implemented system obtains measurement data for a duration of time from an inertial measurement unit (IMU) onboard an underwater device, generates model input data based on at least the measurement data obtained at the plurality of time points, and processes the model input data to generate model output data indicating one or more wave properties using a machine-learning model. The system further determines, based on at least the one or more wave properties, whether the device is safe to be deployed.

Abstract: Techniques are disclosed for generating anomaly scores for a neuro-linguistic model of input data obtained from one or more sources. According to one embodiment, generating anomaly scores includes receiving a stream of symbols generated from an ordered stream of normalized vectors generated from input data received from one or more sensor devices during a first time period. Upon receiving the stream of symbols, generating a set of words based on an occurrence of groups of symbols from the stream of symbols, determining a number of previous occurrences of a first word of the set of words, determining a number of previous occurrences of words of a same length as the first word, and determining a first anomaly score based on the number of previous occurrences of the first word and the number of previous occurrences of words of the same length as the first word.