International Business Machines Corporation

Debasis Ganguly

Pierpaolo Tommasi

Yufang Hou

Martin Gleize

Francesca Bonin

Léa Deleris

Martin Stephenson

Killian Levacher

Ambrish Rawat

Charles Arthur Jochim

Alessandra Pascale

Lea Deleris

Stephane Deparis

Embodiments for providing enhanced data anonymity protection by a processor are disclosed. Selected portions of data intended for distribution in a communication channel or currently distributed on one or more data sources having a potential for revealing identify of a user in a public domain (and/or private domain) may be identified, where an assessment is provided indicating a current status of an amount of data currently exposing the identity of the user in the public domain. The selected portions of the data may be transformed into anonymous data by applying a one or more data corrective operations to prevent further exposure of the identity of the user into the public domain.

Data anonymity protector

Submitting a data space to a provider can risk exposing sensitive information in the data space. To protect the sensitive information, the data space can be analyzed to identify features that are relevant to sensitive information and features that are relevant to a goal task. Features that are relevant to the sensitive information but are not relevant to the goal task can be pruned from the data space, and the pruned data space can be transmitted to the provider.

Sensitive-data-aware encoding

One or more embodiments described herein facilitate identification and mitigation of cognitive bias in data-driven models. In one embodiment, a deep-learning system can comprise a memory that stores computer executable components; and a processor that executes the computer executable components stored in the memory. The computer executable components can comprise: an input component that receives data comprising primary task labels, secondary-identity attributes and a number of potential categories for one or more of the secondary-identity attributes; a machine-learning model that generates one or more predictions based on the received data; and a multi-objective learning component that trains the machine-learning model to mitigate bias from the one or more predictions.

Self-monitoring cognitive bias mitigator in predictive systems

Methods, computer program products and/or systems are provided that perform the following operations: obtaining a collection of objects; constructing a weighted graph based on the collection of objections, wherein the weighted graph preserves neighborhood semantics of objects of the collection of objects; generating partitions of nodes in the weighted graph of a fixed maximum size utilizing combinatorial partitioning; generating a vector for each node based on the partitions of nodes in the weighted graph; determining vector representations for objects in this collection and eventually applying this vector representation of the objects to gain efficiency (e.g., in terms of computation time and memory requirements) for use in downstream tasks such as recommendation.

Discrete representation learning

This method is a process that improves the execution time and maintains very precise clustering effectiveness utilizing a unique algorithm (identified as PPK means) that optimizes a process that is referred to as K-means clustering. The PPK means algorithm utilizes estimation values of signatures of new centroids for speed improvement and encoded data to provide a level of privacy protection. A system comprises a processor, operably coupled to memory, and that executes the computer-executable components, wherein the computer-executable components comprise: an encoding component that encodes a set of real-time valued vectors as bit vectors; and a clustering component that performs K-means clustering on the bit encoded vectors.

Privacy-preserving fast approximate K-means clustering with hamming vectors

A method for automatically detecting inappropriate content on a computing application is provided. The method may include, in response to detecting an attempt to post content on the computing application, parsing the content to identify parts of the content. The method may further include determining whether the parts of the content comprises inappropriate content by applying algorithms to the parts of the content based on information associated with one or more computing applications and by predicting whether potential viewers of the one or more parts of the content view the one or more parts as inappropriate. The method may further include, in response to determining that the one or more parts of the content includes inappropriate content based on the applied algorithms and the potential viewers, generating and displaying feedback and providing suggestions for editing the inappropriate content on the attempted post.

Computer application content detection and feedback

Various embodiments are provided for providing intelligent library management in a computing environment by a processor. Application compatibility may be learned from a plurality of projects, repositories, application libraries, data sources, or a combination thereof. A list of recommended application libraries, ordered according to the application compatibility, may be automatically provided for implementation, integration, or replacements of one or more sections of an application library.

Intelligent application library management

Various embodiments are provided for implementing an approximation nearest neighbour (ANN) search in a computing environment are provided. An approximation nearest neighbour (ANN) of a plurality of feature vectors in hyper-planes with dynamically variable subspaces by searching an inverted index may be retrieved.

Providing approximate top-k nearest neighbours using an inverted list

Tailoring textual content to a target audience by receiving an input of a user, wherein the input of the user includes textual data, identifying a target audience of the textual data based at least in part on the input of the user, determining a style of the target audience, wherein the style is a variety of language used by the target audience, generating a modification recommendation to the textual data of the input of the user based at least in part on the textual data and the determined style.

Content tailoring for diverse audiences

A data structure is used to configure and transform a computer machine learning system. The data structure has one or more records where each record is a (vector) representation of a selected object in a corpus. One or more non-zero parameters in the records define the selected object and the number of the non-zero parameters define a word length of the record. One or more zero-value parameters are in one or more of the records. The word length of the object representation varies, e.g. can increase, as necessary to accurately represent the object within one or more contexts provided during training of a neural network used to create the database, e.g. as more and more contexts are introduced during the training. A minimum number of non-zero parameters are needed and zero-value parameters can be clustered together and compressed to save large amounts of system storage and shorten execution times.

Variable-length word embedding

An information retrieval response may be augmented, based upon a query, with a plurality of selected causality data relating to the query. The information retrieval response may be generated from an information retrieval system.

Providing causality augmented information responses in a computing environment

Embodiments for extraction and summarization of decision discussions of a communication by a processor. The decision elements may be grouped together according to similar characteristics. The decision elements may be linked, and sentiments of the discussion participants towards each of the decision elements may be analyzed. A summary of the plurality of the decision elements may be provided via an interactive graphical user interface (GUI) on one or more Internet of Things (IoT) devices. The summary of the decision elements may be linked to domain knowledge. The summary may be enhanced using a domain knowledge.

Growing community of inventors

Dublin, Ireland

Debasis Ganguly