International Business Machines Corporation

University of Illinois

Rensselaer Polytechnic Institute

Lingfei Wu

Michael John Witbrock

Kun Xu

Ruchi Mahindru

Yu Deng

Liana Liyow Fong

Chia-Yu Chen

Pin-Yu Chen

Chen Wang

Mohammed J Zaki

Rajarshi Haldar

Wei Zhang

Yu Cheng Chen

Jinjun Xiong

Dakuo Wang

Wei Tan

Jing James Xu

Chuang Gan

Vadim Sheinin

Shu Tao

Sijia Liu

Tengfei Ma

Sinem Guven Kaya

Wen-mei William Hwu

Suma Bhat

Yi Wang

Hongyu Gong

Zhiguo Wang

Xuye Liu

Julia Constanze Hockenmaier

Bei Chen

Achille Belly Fokoue-Nkoutche

Kavitha Srinivas

Yada Zhu

Berthold Reinwald

Xue Ying Zhang

Si Er Han

Ji Hui Yang

Lei Yu

Jun Wang

Jinfeng Yi

Tian Gao

Ibrahim Abdelaziz

Cristina Cornelio

Bassem Makni

Xiang Ni

Shen Li

Veronika Thost

Maxwell Crouse

Xiaojie Guo

Meng Jiang

Qingkai Zeng

Wenhao Yu

Cao Xiao

Yu Chen

Nasrullah Sheikh

Xiao Qin

A graph neural network (GNN) training method, system, and computer program product in a graph, include generating, by the computing device, one or more one or more hypothetical edges between two or more nodes of a plurality of nodes of a graph neural network, testing, by the computing device, to determine whether the one or more generated hypothetical edges should be connected by using negative sampling, and permanently connecting, by the computing device, the one or more tested hypothetical edges if the negative sampling indicates the connectivity.

Adaptive self-adversarial negative sampling for graph neural network training

An initial noisy graph topology is obtained and an initial adjacency matrix is generated by a similarity learning component using similarity learning and a similarity metric function. An updated adjacency matrix with node embeddings is produced from the initial adjacency matrix using a graph neural network (GNN). The node embeddings are fed back to revise the similarity learning component. The generating, producing, and feeding back operations are repeated for a plurality of iterations.

Iterative deep graph learning for graph neural networks

Methods and systems for log message aggregation include determining a first similarity distance score for a first incoming message by comparing the first incoming message to one or more stored templates. It is determined that the first incoming message imperfectly matches a matched template of the one or more stored templates, based on the first similarity distance score. A token in the imperfectly matched template is replaced with a wildcard, to reduce the first similarity distance score.

Automatic online log template mining

A computer-implemented method for visual question generation includes training an alignment module to analyze an image, an answer hint, and a visual hint with respect to the image. A k-nearest neighbors (KNN) graph is constructed by performing an aligned embedding for each region of the image. A node embedding component is generated by using a graph embedding component of the KNN graph. A visual question is generated by sequence decoding each image and graph of the image.

Visual question generation with answer-awareness and region-reference

Techniques regarding identifying candidate knowledge graph subgraphs in a question answering over knowledge graph task are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can comprise a question answering over knowledge graph component that encodes graph structure information of a knowledge graph subgraph and a question graph into neural network embeddings.

Querying knowledge graphs with sub-graph matching networks

An adversarial robustness testing method, system, and computer program product include testing, via an accelerator, a robustness of a black-box system under different access settings, where the testing includes tearing down the robustness testing to a subtask of a predetermined size.

Testing adversarial robustness of systems with limited access

Computer-implemented techniques for unsupervised event extraction are provided. In one instance, a computer implemented method can include parsing, by a system operatively coupled to a processor, unstructured text comprising event information to identify candidate event components. The computer implemented method can further include employing, by the system, one or more unsupervised machine learning techniques to generate structured event information defining events represented in the unstructured text based on the candidate event components.

Unsupervised event extraction

A method (and structure and computer product) of machine translation for processing input questions includes receiving, in a processor on a computer, an input question presented in a natural language. The input question is preprocessed to find one or more condition values for possible Structured Query Language (SQL) queries. One or more possible SQL queries are enumerated based on the one or more found condition values and a paraphrasing model is used to rank the enumerated SQL queries. The highest ranked SQL query is executed against a relational database to search for a response to the input question.

Natural language to structured query generation via paraphrasing

Obtain, at a computing device, a segment of computer code. With a classification module of a machine learning system executing on the computing device, determine a required annotation category for the segment of computer code. With an annotation generation module of the machine learning system executing on the computing device, generate a natural language annotation of the segment of computer code based on the segment of computer code and the required annotation category. Provide the natural language annotation to a user interface for display adjacent the segment of computer code.

Learning-based automated machine learning code annotation with graph neural network

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

An adversarial robustness testing method, system, and computer program product include testing a robustness of a black-box system under different access settings via an accelerator.

Embodiments of the present invention provide a computer-implemented method for performing unsupervised feature representation learning for text data. The method generates reference text data having a set of random text sequences, in which each text sequence of set of random text sequences is of a random length and comprises a number of random words, and in which each random length is sampled from a minimum length to a maximum length. The random words of each text sequence in the set are drawn from a distribution. The method generates a feature matrix for raw text data based at least in part on a set of computed distances between the set of random text sequences and the raw text data. The method provides the feature matrix as an input to one or more machine learning models.

Text data representation learning using random document embedding

For a passage text and a corresponding answer text, perform a word-level soft alignment to obtain contextualized passage embeddings and contextualized answer embeddings, and a hidden level soft alignment on the contextualized passage embeddings and the contextualized answer embeddings to obtain a passage embedding matrix. Construct a passage graph of the passage text based on the passage embedding matrix, and apply a bidirectional gated graph neural network to the passage graph until a final state embedding is determined, during which intermediate node embeddings are fused from both incoming and outgoing edges. Obtain a graph-level embedding from the final state embedding, and decode the final state embedding to generate an output sequence word-by-word. Train a machine learning model to generate at least one question corresponding to the passage text and the answer text, by evaluating the output sequence with a hybrid evaluator combining cross-entropy evaluation and reinforcement learning evaluation.

Natural question generation via reinforcement learning based graph-to-sequence model

A computer-implemented method for calculating a similarity between a pair of graph-structured objects by learning-based techniques. The operations include computing the node embeddings of a pair of graph-structured objects of two computer graphs utilizing a hierarchical graph matching network (HGMN). A first component of the HGMN performs graph matching of global-level graph interactions of the two computer graphs. A second component of the HGMN performs graph matching of cross-level node-graph interactions of the two computer graphs. There is an aggregating of features learned from the graph matching of the global-level graph interactions and the cross-level node-graph interactions. At least one of a graph-graph classification or a graph-graph regression is performed utilizing the learned features of the two computer graphs.

Computing graph similarity via graph matching

An approach for a non-factoid question answering framework across tasks and domains may be provided. The approach may include training a multi-task joint learning model in a general domain. The approach may also include initializing the multi-task joint learning model in a specific target domain. The approach may include tuning the joint learning model in the target domain. The approach may include determining which task of the multiple tasks is more difficult for the multi-task joint learning model to learn. The approach may also include dynamically adjusting the weights of the multi-task joint learning model, allowing the model to concentrate on learning the more difficult learning task.

Non-factoid question answering across tasks and domains

Growing community of inventors

Elmsford, NY, United States of America

Lingfei Wu