International Business Machines Corporation

Boaz Carmeli

George Kour

Zeev Waks

Gidon Gershinsky

Ofer Lavi

John Justin Duigenan

Ateret Anaby Tavor

Einat Kermany

Guy Hadash

Ajay K Royyuru

Filippo Utro

Kahn Rhrissorrakrai

Naama Tepper

Guy Lev

Esther Goldbraich

Naama Zwerdling

Takahiko Koyama

Omer Weissbrod

Bilal Erhan

Haggai Roitman

Stephen James Todd

Julian Satran

Michael D Elder

David Konopnicki

Graham Derek Wallis

Noam Slonim

Yosi Mass

Ateret Anaby-Tavor

Yossi Mesika

Amir Kantor

Oren Sar-Shalom

Ruty Rinott

Benjamin M Cohen

Segev Shlomov

Roni Ram

Haim Nelken

Tzilla Eshel

Eyal Ben-David

Matan Vetzler

Koren Ran Lazar

Elad Mezuman

Michael Damein Elder

A computer-implemented method, system and computer program product for identifying a semantic representation for a set of texts. The set of texts is encoded using a language model to obtain a set of corresponding texts' contextualized embeddings. A centroid of the contextualized embeddings is then calculated. A user-designated number of words from a pre-defined vocabulary of the language model that have their non-contextualized embeddings closest to the centroid are then identified. Furthermore, permutations of the identified words are calculated using an n-gram range. The permutations are then encoded using the language model in a contextualized manner. The encoded permutation from the encoded permutations with the greatest similarity to the centroid is identified. The identified encoded permutation is then assigned as the semantic representation of the set of texts. In this manner, the semantic representation is more effectively identified by expanding the embedding space upon which the semantic representation is chosen.

Identifying a semantic representation for texts by expanding the embedding space

Methods and systems for a reinforcement learning system. A spatial and temporal representation of an observed state of an environment is encoded. A previous state is estimated from a given state and a size of a reward is adjusted based on a difference between the estimated previous state and the previous state.

Removing unnecessary history from reinforcement learning state

Balancing an imbalanced dataset, by: Receiving a balancing policy and the imbalanced dataset. Performing initial adjustment of the imbalanced dataset to comply with the balancing policy, by: oversampling one or more underrepresented classes, and, if one or more of the classes are overrepresented, undersampling them. Operating a generative machine learning model to generate samples for the one or more underrepresented classes, based on the initially-adjusted dataset. Operating a machine learning classification model to label the generated samples with class labels corresponding to the one or more underrepresented classes. Selecting some of the generated samples which, according to the labeling, have a relatively high probability of preserving their class labels. Composing a balanced dataset which complies with the balancing policy and comprises: the samples belonging to the one or more underrepresented classes, the selected generated samples, and an undersampling of the samples belonging to the one or more overrepresented classes.

Dataset balancing via quality-controlled sample generation

An example system includes a processor to receive a query. The processor can retrieve ranked candidates from an index based on the query. The processor can re-rank the ranked candidates using a Bidirectional Encoder Representations from Transformers (BERT) query-question (Q-q) model trained to match queries to questions of a frequently asked question (FAQ) dataset, wherein the BERT Q-q model is fine-tuned using paraphrases generated for the questions in the FAQ dataset. The processor can return the re-ranked candidates in response to the query.

Frequently asked questions and document retrieval using bidirectional encoder representations from transformers (BERT) model trained on generated paraphrases

During end-to-end training of a Deep Neural Network (DNN), a differentiable estimator subnetwork is operated to estimate a functionality of an external software application. Then, during inference by the trained DNN, the differentiable estimator subnetwork is replaced with the functionality of the external software application, by enabling API communication between the DNN and the external software application.

Integration of external applications into deep neural networks

Embodiments may determine user intent in conversations with dialogue systems so as to improve the quality of such conversations and to reduce the number of failed conversations. For example, a method may comprise receiving, at a dialogue system, a first text utterance from a user, generating a plurality of second text utterances at the dialogue system in response to the received text utterance, generating a third text utterance based on each generated second text utterance using a trained deep neural network model, generating a score indicating a quality of each conversation, wherein each conversation includes the first text utterance, one of the second text utterances, and the third text utterance based on the one of the second text utterances, and outputting to the user the second text utterance included in the conversation having the highest quality score.

User utterance generation for counterfactual analysis and improved conversation flow

Embodiments of the present invention provide methods, computer program products, and systems for classification and identification of cancer genes while correcting for sample bias for tumor-derived genomic features as well as other biased features using machine learning techniques. Embodiments of the present invention can be used to receive a set of genes that include a first gene and a subset of synthetic genes that include similar features to the first gene and receive a set of gene labels associated with physiological characteristics. Embodiments of the present invention can estimate probabilities that genes in the set of genes are associated with gene labels in the set of gene labels using a machine learning classifier and estimate an effective probability range for the first gene and each gene label based, at least in part, on the first gene's estimated probabilities and the estimated probabilities of one or more of the synthetic genes.

Classification and identification of disease genes using biased feature correction

Embodiments of the present systems and methods may provide techniques for augmenting textual data that may be used for textual classification tasks. Embodiments of such techniques may provide the capability to synthesize labeled data to improve text classification tasks. Embodiments may be specifically useful when only a small amount of data is available, and provide improved performance in such cases. For example, in an embodiment, a method implemented in a computer system may comprise a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor, and the method may comprise fine-tuning a language model using a training dataset, synthesizing a plurality of samples using the fine-tuned language model, filtering the plurality of synthesized samples, and generating an augmented training dataset comprising the training dataset and the filtered plurality of synthesized sentences.

Language-model-based data augmentation method for textual classification tasks with little data

In some examples, a system for executing natural language processing techniques can include a processor to detect text comprising a word and a number. The processor can also embed, via a word embedding model, the word into a first vector of a vector space and embed the number by converting the number into a second vector of the vector space. Additionally, the processor can train a deep neural network to execute instructions based on the first embedded vector of the word and the second embedded vector of the number. Furthermore, the processor can process an instruction based on the trained deep neural network.

Numerical representation in natural language processing techniques

An example system includes a processor to receive a plurality of object aspects of an object to be evaluated using a process, a structure of the process, a plurality of extracted facts from documents, a tree related to the plurality of object aspects and the structure, and a thesis for each leaf in the tree. The processor is also to relate the extracted facts to the theses in the tree. The processor is to generate a score for each leaf corresponding to a fact in the tree. The processor is to generate a thesis score and a thesis summary for each thesis based on the scores and the summaries of related facts for each thesis. The processor is to further generate a final score for the object based on the thesis scores.

Evaluating theses using tree structures

Embodiments are directed to computer implemented method of assessing a relevancy of a pathway to a disease of interest, the pathway having a source and a target. The method includes developing an impact of the source on the pathway. The method further includes developing a value of targeting, based at least in part on an alteration of the pathway, the pathway with a drug of interest. The method further includes identifying a relationship between the source and the target within the pathway. The method further includes combining: the impact of the source on the pathway; the value of targeting, based at least in part on the alteration of the pathway, the pathway with a drug of interest; and the relationship between the source and the target within the pathway, wherein the combining results in an assessment that represents the relevancy of the pathway to the disease of interest.

Simplified visualization and relevancy assessment of biological pathways

A computer implemented method, computerized apparatus and computer program product for inconsistency detection between structured and non-structured data. The method comprising: automatically determining, by a computer, inconsistencies between fields in electronics records, the fields comprise at least a structured field and a non-structured field, the fields are designed to be able to include overlapping information in structured and non-structured form; and indicating, by the computer, to a user potential inconsistencies. Optionally, the indication uses a visual cue when displaying the electronic record to the user, wherein the visual cue indicates the fields which are determined to comprise inconsistent content.

Inconsistency detection between structured and non-structured data

A multicast host for communicating information published about any one of a set of topics to one or more authorised subscribers to those topics, the set of topics being partitioned into one or more partition elements, each partition element having a partition element encryption key associated therewith, wherein each of the one or more partition elements is a disjoint proper subset of the set of topics, the host comprising: means for receiving information relating to a topic; means for determining a partition element for the topic; means for retrieving a partition element encryption key associated with the partition element; means for encrypting the information with the retrieved partition element encryption key; and means for communicating the information to the one or more authorised subscribers.

Access control over multicast

Growing community of inventors

Koranit, Israel

Boaz Carmeli