International Business Machines Corporation

Globalfoundries Inc.

University of Illinois

Rensselaer Polytechnic Institute

Yada Zhu

Xuan Liu

Jayant R Kalagnanam

Ajay Ashok Deshpande

Brian Leo Quanz

Ali Koc

Yingjie Li

Arun Hampapur

Robert J Baseman

Emmanuel Yashchin

Kumar Bhaskaran

Jingrui He

Lei Cao

Pin-Yu Chen

Wei Zhang

Xiaodong Cui

Nitin Gaur

Anna Wanda Topol

Jinjun Xiong

Hongfei Li

Jae-Eun Park

Amit Dhurandhar

Zhiguo Li

Jonathan R Hosking

Stuart A Siegel

Jonathan R M Hosking

Guangnan Ye

Chandrasekhar Narayanaswami

Young Min Lee

Chia-Yu Chen

Petr Novotny

Lingfei Wu

Yacov Manevich

Yang Zhang

Rong Nickle Chang

Christopher Scott Milite

Artem Barger

Geert Janssen

Lucian Popa

Sijia Liu

David E Cox

Lexing Xie

Upendra D Chitnis

Youngdeok Hwang

Bert Huang

Bo Li

Giovanni Mariani

Koushik Kar

Di Chen

Yunan Ye

Aparna Gupta

Sai Radhakrishna Manikant Sarma Palepu

Zixuan Yuan

A computer implemented method and system of evaluating a fulfillment strategy in an omnichannel distribution system is provided. Input parameters are received from a computing device of a user. Historical data related to a network of nodes is received from a data repository. A synthetic demand status is determined based on the historical data and the input parameters. A synthetic network status based on the historical data and the input parameters are determined. A fulfillment strategy is identified based on the synthetic demand status and the synthetic network status. Key performance indicators (KPIs) for the fulfillment strategy are determined based on the synthetic demand status and the synthetic network status.

System for modeling the performance of fulfilment machines

A graph neural network can be built and trained to predict a risk of an entity. A multi-relational graph network can include a first graph network and a second graph network. The first graph network can include a first set of nodes and a first set of edges connecting some of the nodes in the first set. The second graph network can include a second set of nodes and a second set of edges connecting some of the nodes in the second set. The first set of nodes and the second set of nodes can represent entities, the first set of edges can represent a first relationship between the entities and the second set of edges can represent a second relationship between the entities. A graph convolutional network (GCN) can be structured to incorporate the multi-relational graph network, and trained to predict a risk associated with a given entity.

Multi-relational graph convolutional network (GCN) in risk prediction

Hessian matrix-free sample-based techniques for model explanations that are faithful to the model are provided. In one aspect, a method for explaining a machine learning model {circumflex over (θ)} (e.g., for natural language processing) is provided. The method includes: training the machine learning model {circumflex over (θ)} with training data D; obtaining a decision of the machine learning model {circumflex over (θ)}; and explaining the decision of the machine learning model {circumflex over (θ)} using training examples from the training data D.

Faithful and efficient sample-based model explanations

A machine learning model can be trained to predict one or more financial indicators using earnings call transcripts augmented with counterfactual information. Using faithful gradient-based method, prediction results with respect to a particular counterfactual information can be explained. Based on the explanation, the counterfactual information determined to have most impact on prediction results can be selected for updating the machine learning model.

Counterfactual neural network learning for contextual enhanced earnings call analysis

Unsupervised cross-domain data augmentation techniques for long-text document based prediction and explanation are provided. In one aspect, a system for long-document based prediction includes: an encoder for creating embeddings of long-document texts with hierarchical sparse self-attention, and making predictions using the embeddings of the long-document texts; and a multi-source counterfactual augmentation module for generating perturbed long-document texts using unlabeled sentences from at least one external source to train the encoder. A method for long-document based prediction is also provided.

Unsupervised cross-domain data augmentation for long-document based prediction and explanation

An example operation may include one or more of receiving, via a custodial service, a request to transact with a digital asset owned by a user and temporarily in custody of the custodial service, generating a blockchain transaction comprising an identifier of the digital asset on a blockchain ledger, an identifier of the custodial service, and an identifier of a recipient of the digital asset, signing the blockchain transaction with a key from a trifocal key which proves that the custodial service is authorized to transact with the digital asset on behalf of the user, and storing the signed blockchain transaction on a blockchain ledger.

Trifocal key for controlling custodians of digital assets

A processor training a reinforcement learning model can include receiving a first dataset representing an observable state in reinforcement learning to train a machine to perform an action. The processor receives a second dataset. Using the second dataset, the processor trains a machine learning classifier to make a prediction about an entity related to the action. The processor extracts an embedding from the trained machine learning classifier, and augments the observable state with the embedding to create an augmented state. Based on the augmented state, the processor trains a reinforcement learning model to learn a policy for performing the action, the policy including a mapping from state space to action space.

State-augmented reinforcement learning

Techniques that facilitate graph similarity analytics are provided. In one example, a system includes an information component and a similarity component. The information component generates a first information index indicative of a first entropy measure for a first graph-structured dataset associated with a machine learning system. The information component also generates a second information index indicative of a second entropy measure for a second graph-structured dataset associated with the machine learning system. The similarity component determines similarity between the first graph-structured dataset and the second graph-structured dataset based on a graph similarity computation associated with the first information index and the second information index.

Graph similarity analytics

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.

Explainable machine learning based on heterogeneous data

A task-based learning using task-directed prediction network can be provided. Training data can be received. Contextual information associated with a task-based criterion can be received. A machine learning model can be trained using the training data. A loss function computed during training of the machine learning model integrates the task-based criterion, and minimizing the loss function during training iterations includes minimizing the task-based criterion.

Task-oriented machine learning and a configurable tool thereof on a computing environment

A deep-learning neural network can be trained to model a probability distribution of the asset-price trends for a future time period using a training data set, which can include asset-price trends of a plurality of assets over a past time period and a latent vector sampled from a prior distribution associated with the asset-price trends of a plurality of assets. The training data set can represent a time series data. A portfolio optimization can be executed on the modeled probability distribution to estimate expected risks and returns for different portfolio diversification options.

Generative network based probabilistic portfolio management

Embodiments herein describe a return network simulation system that can simulate changes in a retailer's return network to determine the impact of those changes. Advantageously, being able to accurately simulate the retailer's return network means changes can be evaluated without first making those adjustments in the physical return network. Doing so avoids the cost of implementing the changes on the return network without first being able to predict whether the changes will have a net positive result (e.g., a positive result that offsets any negative results). A retailer can first simulate the change on the return network, review how the change affects one or more KPIs, and then decide whether to implement the change in the actual return network. As a result, the retailer has a reliable indicator whether the changes will result in a desired effect.

Simulating a return network

A computer implemented method and system of setting values of parameters of nodes in an omnichannel distribution system, the method comprising is provided. Input parameters are received from a computing device. Historical data related to the network of nodes is received from a data repository. A synthetic scenario is determined based on the received input parameters and the historical data. Each node is clustered into a corresponding category. For each category of nodes, key parameters are identified. A range of each key parameter is determined based on the synthetic scenario. A number of simulations N to perform with data sampled from the synthetic scenario within the determined range of each key parameter is determined. For each of the N simulations, a multi-objective optimization is performed to determine a cost factor of the parameter settings. The parameter settings with a lowest cost factor are selected.

Fulfilment machine for optimizing shipping

A computer implemented method and system of calculating labor resources for a network of nodes in an omnichannel distribution system. Input parameters are received from a computing device of a user. Historical data related to a network of nodes is received, from a data repository. A synthetic scenario is determined based on the received input parameters and the historical data. For each node, key parameters are identified and set based on a multi-objective optimization, wherein the multi-objective optimization includes a synthetic inventory allocation to the node based on the synthetic scenario. A synthetic labor efficiency is determined for the node from the synthetic scenario. Labor resources are calculated based on the synthetic inventory allocation for the synthetic scenario. The labor resources of at least one node are displayed on a user interface of a user device.

Machine for labor optimization for efficient shipping

Embodiments describing an approach to aligning multiple time series, calculating an indicator function, estimating a coefficient vector based on the indicator function, and updating the coefficient vector. Additionally, embodiments comprise determining if a change in the coefficient vector is less than a predetermined value, and responsive to determining the change in the coefficient vector is less than the predetermined value outputting a target time series for controlling aluminum smelting.

Manufacturing process control based on multimodality and multi-resolution time series data

Growing community of inventors

White Plains, NY, United States of America

Yada Zhu