Abstract: In an embodiment, a method for generating contrastive information for a classifier prediction comprises receiving image data representative of an input image, using a deep learning classifier model to predict a first classification for the input image, evaluating the input image using a plurality of classifier functions corresponding to respective high-level features to identify one or more of the high-level features absent from the input image, and identifying, from among the high-level features absent from the input image, a pertinent-negative feature that, if added to the input image, will result in the deep learning classifier model predicting a second classification for the modified input image, the second classification being different from the first classification. In an embodiment, the method includes creating a pertinent-positive image that is a modified version of the input image that has the first classification and fewer than all superpixels of the input image.

Abstract: A multi-phase inductor structure is provided. The multi-phase inductor structure includes a first magnetic core, two second magnetic cores, and two first electrical conductors. The two second magnetic cores are respectively arranged on opposite sides of the first magnetic core, and each have a first engagement surface. A first annular convex wall and a first upright convex wall are formed on the first engagement surface, and a first recess is formed therebetween. The two first electrical conductors are respectively arranged in two of the first recesses of the first engagement surface, and each have has a first body and two first pins that are respectively connected to two ends of the first body. The two first pins extend in opposite directions. A magnetic permeability of the first magnetic core is different from a magnetic permeability of each of the two second magnetic cores.

Abstract: A reinforcement learning system includes a plurality of agents, each agent having an individual reward function and one or more safety constraints that involve joint actions of the agents, wherein each agent maximizes a team-average long-term return in performing the joint actions, subject to the safety constraints, and participates in operating a physical system. A peer-to-peer communication network is configured to connect the plurality of agents. A distributed constrained Markov decision process (D-CMDP) model is implemented over the peer-to-peer communication network and is configured to perform policy optimization using a decentralized policy gradient (PG) method, wherein the participation of each agent in operating the physical system is based on the D-CMDP model.

Abstract: A certification method, system, and computer program product include certifying an adversarial robustness of a convolutional neural network by deriving an analytic solution for a neural network output using an efficient upper bound and an efficient lower bound on an activation function and applying the analytic solution in computing a certified robustness.

Abstract: Methods and systems for explainable machine learning are described. In an example, a processor can receive a data set from a plurality of data sources corresponding to a plurality of domains. The processor can train a machine learning model to learn a plurality of vectors that indicate impact of the plurality of domains on a plurality of assets. The machine learning model can be operable to generate forecasts relating to performance metrics of the plurality of assets based on the plurality of vectors. In some examples, the machine learning model can be a neural attention network with shared hidden layers.

Abstract: Methods, systems, and circuits for training a neural network include applying noise to a set of training data across wordlines using a respective noise switch on each wordline. A neural network is trained using the noise-applied training data to generate a classifier that is robust against adversarial training.

Abstract: Techniques for sanitization of machine learning (ML) models are provided. A first ML model is received, along with clean training data. A path is trained between the first ML model and a second ML model using the clean training data. A sanitized ML model is generated based on at least one point on the trained path. One or more ML functionalities are then facilitated using the sanitized ML model.

Abstract: Embodiments relate to a system, program product, and method for inducing creativity in an artificial neural network (ANN) having an encoder and decoder. Neurons are automatically selected and manipulated from one or more layers of the encoder. An encoded vector is sampled for an encoded image. Decoder neurons and a corresponding activation pattern are evaluated with respect to the encoded image. The decoder neurons that correspond to the activation pattern are selected, and an activation setting of the selected decoder neurons is changed. One or more novel data instances are automatically generated from an original latent space of the selectively changed decoder neurons.

Abstract: The present invention discloses a method for determining a maximum value of a heart activity parameter of a user performing a physical activity. Acquire first heart activity data in a first duration of the physical activity performed by the user. Acquire motion data in the first duration of the physical activity performed by the user. Calculate second heart activity data based on the motion data in the first duration of the physical activity performed by the user by a mathematical model and estimate the maximum value of the heart activity parameter of the user based on a comparison between the first heart activity data and the second heart activity data.

Abstract: Machine learning-based techniques for summarizing collections of data such as image and video data leveraging side information obtained from related (e.g., video) data are provided. In one aspect, a method for video summarization includes: obtaining related videos having content related to a target video; and creating a summary of the target video using information provided by the target video and side information provided by the related videos to select portions of the target video to include in the summary. The side information can include video data, still image data, text, comments, natural language descriptions, and combinations thereof.

Abstract: A system and method for designing protein sequences conditioned on a specific target fold. The system is a transformer-based generative framework for modeling a complex sequence-structure relationship. To mitigate the heterogeneity between the sequence domain and the fold domain, a Fold-to-Sequence model jointly learns a sequence embedding using a transformer and a fold embedding from the density of secondary structural elements in 3D voxels. The joint sequence-fold representation through novel intra-domain and cross-domain losses with an intra-domain loss forcing two semantically similar (where the proteins should have the same fold(s)) samples from the same domain to be close to each other in a latent space, while a cross-domain loss forces two semantically similar samples in different domains to be closer. In an embodiment, the Fold-to-Sequence model performs design tasks that include low resolution structures, structures with region of missing residues, and NMR structural ensembles.

Abstract: A method, system, and computer program product, including generating a contrastive explanation for a decision of a classifier trained on structured data, highlighting an important feature that justifies the decision, and determining a minimal set of new values for features that alter the decision.

Abstract: Training a machine learning model can include receiving time series data. A knowledge graph structure can be received including nodes and edges, the nodes representing entities associated with the time series data, the edges representing relationships between the nodes connected by the edges. A machine learning model can be structured to forecast a prediction using the time series data. The machine learning model can be structured to integrate the knowledge graph structure as an error term in the machine learning model. The machine learning model can be trained to forecast the prediction based on the time series data and the knowledge graph structure. The error term representing the knowledge graph structure can be regularized for sparsity during training.

Abstract: Scalable distributed adversarial training techniques for robust deep neural networks are provided. In one aspect, a method for adversarial training of a deep neural network-based model by distributed computing machines M includes, by distributed computing machines M: obtaining adversarial perturbation-modified training examples for samples in a local dataset D(i); computing gradients of a local cost function fi with respect to parameters ? of the deep neural network-based model using the adversarial perturbation-modified training examples; transmitting the gradients of the local cost function fi to a server which aggregates the gradients of the local cost function fi and transmits an aggregated gradient to the distributed computing machines M; and updating the parameters ? of the deep neural network-based model stored at each of the distributed computing machines M based on the aggregated gradient received from the server.

Abstract: Techniques for training robust machine learning models for adversarial input data. Training data for a machine learning (ML) model is received. The training data includes a plurality of labels for data elements. First modified training data is generated by modifying one or more of the plurality of labels in the training data using parameterized label smoothing with a first optimization parameter. The ML model is trained using the first modified training data.

Abstract: A trained machine learning model and a training dataset used to train the trained machine learning model can be received. Based on the training dataset, unsupervised adversarial examples can be generated. Robustness of the trained machine learning model can be determined using the generated unsupervised adversarial examples. The training dataset can be augmented with the generated unsupervised adversarial examples. The trained machine learning model can be retrained using the augmented training dataset.

Abstract: Generate, for each of the words of a common vocabulary of first and second text corpora, a first word embedding vector in the first text corpus and a second word embedding vector in the second text corpus. Generate, for each word in a random sample of non-landmark words, an artificially shifted word embedding vector by modifying the first word embedding vector for that word. Train a machine learning classifier to predict whether an artificial shift has been injected for a given word, based on the artificially shifted word embedding vector and the second word embedding vector for the given word. Predict semantic shifts for at least a plurality of the words of the common vocabulary by providing the first word embedding vectors and the second word embedding vectors for at least the plurality of the words of the common vocabulary as input to the trained machine learning classifier.

Abstract: Embodiments relate to a system, program product, and method to support a convolutional neural network (CNN). A class-specific discriminative image region is localized to interpret a prediction of a CNN and to apply a class activation map (CAM) function to received input data. First and second attacks are generated on the CNN with respect to the received input data. The first attack generates first perturbed data and a corresponding first CAM, and the second attack generates second perturbed data and a corresponding second CAM. An interpretability discrepancy is measured to quantify one or more differences between the first CAM and the second CAM. The measured interpretability discrepancy is applied to the CNN. The application is a response to an inconsistency between the first CAM and the second CAM and functions to strengthen the CNN against an adversarial attack.

Abstract: One or more computer processors generate a plurality of adversarial perturbations associated with a model, wherein the plurality of adversarial perturbations comprises a universal perturbation and one or more per-sample perturbations. The one or more computer processors identify a plurality of neuron activations associated with the model and the plurality of generated adversarial perturbations. The one or more computer processors maximize the identified plurality of neuron activations. The one or more computer processors determine the model is a Trojan model by leveraging one or more similarities associated with the maximized neuron activations and the generated adversarial perturbations.

Abstract: An inductor and a power module are respectively provided. The inductor includes an insulating body and a conductive body. The insulating body has a top surface and a bottom surface. The conductive body includes two pin parts and a heat dissipation part. A portion of each of the pin parts is exposed outside the bottom surface. The portions of the two pin parts exposed outside the insulating body are configured to fix to a circuit board. The heat dissipation part is connected to the two pin parts, the heat dissipation part is exposed outside the top surface, and the heat dissipation part is configured to connect to an external heat dissipation member. When the inductor is fixed to the circuit board through the two pin parts exposed outside the bottom surface, the two pin parts and the bottom surface jointly define an accommodating space for accommodating a chip.