Abstract: A system, method and computer program code are described that provide a computer-implemented method for characterizing and evaluating a pre-trained model. The method includes training an unsupervised model using non-parametric property-driven subset scanning to provide a characterization of embeddings of the pre-trained model and using the characterizations to compute a metric to contrast different domain adaptation methods for the pre-trained model.

Abstract: Anomaly detection in neural networks is provided. The method comprises extracting, from different layers of a recurrent neural network (RNN) for a specified time interval, a number of on-the-fly node activations produced by sequences of event data from a number of data sources and sensors. The method references known node activations of the RNN produced by normal sequences of event data, and for each layer of the RNN, calculates a maximum nonparametric divergence of the on-the-fly node activations from the known node activations. For each layer of the RNN, the method determines a subset of nodes that most contribute to the maximum nonparametric divergence for that layer for a given time window and identifies data sources or sensors from among the number of data sources and sensors responsible for activating the subset of nodes that most contribute to the maximum nonparametric divergence for each layer.

Abstract: A processor may receive input data. The processor may train a first machine learning model with the input data to extract a group of activation nodes. The processor may extract the group of activation nodes. The processor may detect anomalous nodes from the group of activation nodes.

Abstract: A present invention embodiment prevents patch-based adversarial attacks. A plurality of time-series images are processed using a machine learning model to identify an anomaly, wherein the anomaly comprises an adversarial patch-based attack, and wherein the anomaly is present in a locality comprising a particular time and space in the plurality of time-series images. A subset of the plurality of time-series images are modified based on the locality of the anomaly to mask the anomaly from detection by a trained image processing model.

Abstract: One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to automated control for a physical system with generic forecasting models. The computer-implemented system can comprise a memory that can store computer executable components. The computer-implemented system can further comprise a processor that can execute the computer executable components stored in the memory, wherein the computer executable components can comprise a discovery component that obtains a candidate anomalous subset of a dataset, a scoring component that computes a diversity score of the candidate anomalous subset relative to selected subsets of the dataset, and a selection component that selects the candidate anomalous subset based on the diversity score. Furthermore, the level of diversity between subsets can be controlled by a user. Moreover, records can be penalized to enable search space exploration and mitigate redundancy at the record level.

Abstract: With a trained, computerized discontinuation predictor machine learning component (MLC), predict, based on an input time series, a time when a subject will discontinue a course of medical treatment; with a trained, computerized pattern behavior extractor MLC, extract from said input time series the top k discriminatory sequences via discriminatory sub-sequence mining (said top k discriminatory sequences differentiate between first and second classes of interest to provide a hypothesis for downstream analysis of a cause of discontinuing said course of treatment). With a trained, causal effect estimator computerized MLC, determine a reason why said subject will discontinue said course of medical treatment, based on said top k discriminatory sequences and additional data; and with a computerized user interface, provide said time and said reason why to a responsible party to initiate an intervention.

Abstract: A computer-implemented method, according to one approach, includes: receiving a new set of test data and evaluating the test data using a pre-trained deep neural network. In response to evaluating the test data, activations are extracted from layers of the deep neural network. Compressed histograms are further used to determine p-values for the extracted activations. The p-values are evaluated and portions of the test data that are determined as being anomalous, based at least in part on the evaluation of the p-values, are retained.

Abstract: Domain-specific properties are extracted from a plurality of datasets related to input molecules and generated molecules. A similarity of a given one of the generated molecules with a candidate molecule is evaluated and the evaluated similarity is aggregated across multiple constraints to generate a single score. An ability of a given generative model to mimic the input molecules is quantified based on an aggregation of scores for multiple properties. The generative models are ranked by their ability to mimic the input molecules and evaluation properties based on the aggregated scores. One or more rates in the generated molecules are quantified based on the aggregated single score. One of the generative models is selected based on the ranking and a new molecule is generated using the selected generative model.

Abstract: Anomaly detection in neural networks is provided. The method comprises extracting, from different layers of a recurrent neural network (RNN) for a specified time interval, a number of on-the-fly node activations produced by sequences of event data from a number of data sources and sensors. The method references known node activations of the RNN produced by normal sequences of event data, and for each layer of the RNN, calculates a maximum nonparametric divergence of the on-the-fly node activations from the known node activations. For each layer of the RNN, the method determines a subset of nodes that most contribute to the maximum nonparametric divergence for that layer for a given time window and identifies data sources or sensors from among the number of data sources and sensors responsible for activating the subset of nodes that most contribute to the maximum nonparametric divergence for each layer.

Abstract: In an approach to improve platform modification by dynamically modifying content based on user feedback. Embodiments suggest one or more solutions and recommendations, via a user interface, to one or more users that comply with one or more standards, regulations, and resources of an organization. Further, embodiments verify the solutions and recommendations that comply with the one or more standards, regulations, and resources of the organization, and automatically generate a prototype solution. Additionally, embodiments provide, by the user interface, one or more augmented solutions for the one or more users, wherein augmented solutions are provided to the one or more users requiring a temporary solution; and label and cluster user feedback and associated solutions of the one or more users to provide effective solutions for future users based on a similarity of the feedback of the one or more users.

Abstract: A present invention embodiment prevents patch-based adversarial attacks. A plurality of time-series images are processed using a machine learning model to identify an anomaly, wherein the anomaly comprises an adversarial patch-based attack, and wherein the anomaly is present in a locality comprising a particular time and space in the plurality of time-series images. A subset of the plurality of time-series images are modified based on the locality of the anomaly to mask the anomaly from detection by a trained image processing model.

Abstract: A supervised discriminator for detecting bio-markers in an audio sample dataset is trained and a denoising autoencoder is trained to learn a latent space that is used to reconstruct an output audio sample with a same fidelity as an input audio sample of the audio sample dataset. A conditional auxiliary generative adversarial network (GAN) trained to generate the output audio sample with the same fidelity as the input audio sample, wherein the output audio sample is void of the bio-markers. The conditional auxiliary generative adversarial network (GAN), the corresponding supervised discriminator, and the corresponding denoising autoencoder are deployed in an audio processing system.

Abstract: An exemplary method includes vehicle-mounted sensors continuously detecting vehicle speed and vehicle tire and steering vibrations; a processor implementing a machine-learning program that continuously monitors signals from the vehicle-mounted sensors and compares detected vehicle tire and steering vibrations to upper bounds corresponding to detected vehicle speed; and the processor alerting a vehicle driver that wheel or tire service is required based on the detected vehicle tire and steering vibrations exceeding the upper bounds. An exemplary apparatus includes a vehicle; tires mounted to the vehicle; a speed sensor mounted to the vehicle; a vibration sensor mounted to the vehicle; and a processor connected in communication with the speed sensor and the vibration sensor. The processor is adapted to implement any of the method steps above.

Abstract: A computer system may create a language model corpus including multilingual alignment for training a combined language model and train (or pre-train) the combined language model. The computer system may create an adverse medication reaction corpus to include adverse medication reaction utterances and label an N-gram of an utterance in the adverse medication reaction utterances as a response to query, for multiple N-grams. The computer system may generate a code-mixed utterance model to perform code-mixed utterances in a turn by turn dialogue, by at least adding additional output layer including at least a start vector, language vector, and a query vector including at least the labeled N-gram, which are additional to the combined language model's predicted next words.

Abstract: Protected features from media that enable bias in a decision-making process of an artificial intelligence (AI) system are identified and ranked. Adversarial artifacts are generated, using a generative adversarial network, to obfuscate at least a portion of the ranked protected features and the adversarial artifacts are integrated into the media.

Abstract: An illustrative embodiment includes a method for detecting whether a given item within input data is anomalous. The method includes: extracting activations at respective nodes of an autoencoder trained on the data, the activations comprising activations at the respective nodes for the given item within the data and for one or more other items within the data; calculating p-values corresponding to the respective nodes, wherein the p-value for a given node represents a proportion of the activations at the given node for the other items which are greater than the activations at the given node for the given item; determining at least one score at least in part by scanning for a subset of the respective nodes which maximizes a statistical scoring function applied to the corresponding p-values; and providing a visualization of at least the given item within the input data to a human user.

Abstract: An exemplary method includes intercepting data in a virtual presence solution; classifying the intercepted data against machine-readable conduct guidelines by processing the intercepted data in a neural network; and intervening in the virtual presence solution in response to a component of the intercepted data being classified to a match with at least one of the machine-readable conduct guidelines.

Abstract: Members of an audience of a visually dynamic event are clustered using a plurality of sources. A current point of interest (POI) of the visually dynamic event and a future POI of the visually dynamic event are identified across the member clusters. An effectiveness score for given content and a given member cluster is computed for the current point of interest and an effectiveness score for a given content and a given member cluster is computed for the future POI by tracking a position, a speed, and a direction of movement of the current POI. A location of a background area is determined and ranked for each member cluster based on the current point of interest. Electronic displays that correspond to the ranked background areas are identified and ranked. Content is distributed across the electronic displays for each current time period based on the ranked display areas.

Abstract: Embodiments of the present invention provide a computer system, a computer program product, and a method that comprises generating a baseline prediction for each database in a plurality of databases, wherein the generated baseline prediction includes datapoints for each performance feature of a respective database in the plurality of databases; generating a linkage matrix using a dimension reduction algorithm in an equal-dimensional feature space based on a measured difference between actual performance datapoints of each respective database and respective datapoints associated with generated baseline prediction for each database in the plurality of databases; linking one or more databases in the plurality of databases based on the measured difference within the generated linkage matrix reaching a predetermined threshold of measured performance; and generating a design using the linked databases for a computer system capable of being displayed through a user interface.

Abstract: A search query can be received. User parameters can be identified based on the search query. The search query can be refined to include the user parameters. A search result from a search for media content using the refined search query can be received. Based on at least one search result received from the search and based on the user parameters, an augmented media content can be generated. Playing of the augmented media content can be synchronized with a user's activity by controlling playing of the augmented media content while detecting the user's activity pace.