Tata Consultancy Services Limited

Indian Statistical Institute

Tanushyam Chattopadhyay

Arpan Pal

Aniruddha Sinha

Prateep Misra

Avik Ghose

Soma Bandyopadhyay

Puneet Kumar Agarwal

Chirabrata Bhaumik

Arijit Ukil

Snehasis Banerjee

Anirban Dutta Choudhury

Utpal Garain

Arijit Sinharay

Ramu Reddy Vempada

Dipti Prasad Mukherjee

Aishwarya Visvanathan

Sangheeta Roy

Suvra Dutta

Shubhrangshu Ghosh

Mohd Bilal Arif Syed

Satanik Panda

Abhisek Das

Bidyut Baran Chaudhuri

Rohan Banerjee

Kingshuk Chakravarty

Vivek Chandel

Ayan Mukherjee

Swagata Biswas

Angshul Majumdar

Ishan Sahu

Arindam Pal

Apurbaa Mallik

Dhiman Chattopadhyay

Tulika Bose

Provat Biswas

Santa Maiti

Plaban Kumar Bhowmick

Utpal Garin

Anurag Kumar

Kallol Ghosh

Tamal Batabyal

Ramu Vempada Reddy

Debabrata Pradhan

Soumali Roychowdhury

Ujjwal Bhattacharya

Arindam Halder

Abhishek Roy Choudhury

Sangram Dasharath Gaikwad

Tania Ghosh

Dibyendu Biswas

Hemanta Dutta

Abhishek Choudhury

Indrajit Bhattachyarya

State-of-the-art approaches have concentrated on building solution(s) to match the amplitude of a time series with a user given one. However, these have failed to implement solution(s) which enables searching for pattern(s) that can depict human vision psychology. Embodiments of the present disclosure determine occurrence of pattern of interest in time series data for anomaly detection, wherein time series data is obtained, and first order derivative is computed. Further an angle of change in direction is derived based on a gradient of change in value of the time series data. This angle is further converted to a measurement unit. The time series data is quantized into bins and a weighted finite state transducers diagram (WFSTD) is obtained based on domain knowledge which is then converted to specific pattern. The specific pattern is searched in the bins to determine occurrence/count of the specific pattern for anomaly detection.

Systems and methods for determining occurrence of pattern of interest in time series data

A method and system for incorporating regression into a Stacked Auto Encoder utilizing deep learning based regression technique that enables joint learning of parameters for a regression model to train the SAE for a regression problem. The method comprises generating a regression model for the SAE for solving the regression problem, wherein regression model is formulated as a non-convex joint optimization function for an asymmetric SAE. The method further comprises reformulating the non-convex joint optimization function as an Augmented Lagrangian formulation in terms of a plurality of proxy variables and a plurality of hyper parameters. The method comprises splitting the Augmented Lagrangian formulation into sub-problems using Alternating Direction Method of Multipliers and jointly learning parameters for the regression model to train the SAE for the regression problem. The learned weights enable estimating the unknown target values.

Method and system for incorporating regression into stacked auto encoder (SAE)

Data cataloging has become a necessity for empowering organizations with analytical ability. Conventional cataloging systems may fail to provide proper visualization of data to the different stakeholders of an organization. The present disclosure provides a hierarchical dynamic cataloging system so that visualization of data at different levels would be possible for different stake holders. In the present disclosure, a hierarchical structure of algorithms and multiple stake holders along with relevant metadata is generated. Further, a catalog is generated by performing a mapping across components comprised in the hierarchical structure and identifying relationship across the components based on mapping. The catalog gets dynamically updated and provides a dynamic view of the algorithms and associated metadata to the multiple stakeholders of an organization. Further, the disclosure supports reuse of already developed algorithms across multiple applications and domains resulting in optimization of resources and time.

Methods and systems for hierarchical dynamic cataloging

Systems and methods for obtaining optimal mother wavelets for facilitating machine learning tasks. The traditional systems and methods provide for selecting a mother wavelet and signal classification using some traditional techniques and methods but none them provide for selecting an optimal mother wavelet to facilitate machine learning tasks. Embodiments of the present disclosure provide for obtaining an optimal mother wavelet to facilitate machine learning tasks by computing values of energy and entropy based upon labelled datasets and a probable set of mother wavelets, computing values of centroids and standard deviations based upon the values of energy and entropy, computing a set of distance values and normalizing the set of distance values and obtaining the optimal mother wavelet based upon the set of distance values for performing a wavelet transform and further facilitating machine learning tasks by classifying or regressing, a new set of signal classes, corresponding to a new set of signals.

Systems and methods for obtaining optimal mother wavelets for facilitating machine learning tasks

Sensor data (or IoT) analytics plays a critical role in taking business decisions for various entities (e.g., organizations, project owners, and the like). However, scaling of such analytical solutions beyond certain point requires adopting to various computing environments which seems to be challenging with the constrained resources available. Embodiments of the present disclosure provide system and method for analysing and executing sensor observational data in computing environments, wherein extract, transform, load (ETL) workflow pipeline created by users in the cloud, can be seamlessly deployed to job execution service available in cloud/edge without any changes in the code/config by end user. The configuration changes are internally handled by the system based on the selected computing environment and queries are executed either in distributed or non-distributed environments to output data frames. The data frames are further pre-processed in a desired computing environment and thereafter visualized accordingly.

Systems and methods for recommending execution environment for analysing sensor observational data

Machine Learning (ML) models are deployed in digital platforms for data analytics. However, it is realized that there is growing trends of recognition that machine learning models expose new vulnerabilities in software systems, for instance training data poisoning, adversarial responses, model extraction, and the like. Embodiments of the present disclosure provide systems and methods for safeguarding training dataset by exploiting immutability feature and generating immutable machine learning models for data analytics. More specifically, immutable records of events are governed by smart contracts within highly secure permissioned distributed ledger. This dataset is used for training multiple machine learning models which are immutable in nature and further utilized for triggering actions for incoming request(s) from IoT platforms.

Immutable records based generation of machine learning models for dynamically triggering actions

A multi-dimensional sensor data analysis system and method is provided. The multi-dimensional sensor data analysis system receives indoor and outdoor location, online and physical activity, online and physical proximity and additional a plurality of inputs (specific to a user), for example, surrounding of the subject, physiological parameters of the subject and recent social status of the subject, both online and offline. The multi-dimensional sensor data analysis system processes these inputs along with the knowledge of past behavior and traditional parameters of location, proximity and activity by performing a multi-dimensional sensor data analysis fusion technique, producing one or more outputs, for example, predicting or determining a human behaviour to a given stimuli.

Multi-dimensional sensor data based human behaviour determination system and method

Development of sensor data based descriptive and prescriptive system involves machine learning tasks like classification and regression. Any such system development requires the involvement of different stake-holders for obtaining features. Such features typically obtained are not interpretable for 1-D sensor signals. Embodiments of the present disclosure provide systems and methods that perform signal analysis for features extraction and interpretation thereof wherein input is raw signal data where origin of a feature is traced to signal data, and mapped to domain/application knowledge. Feature(s) are extracted using deep learning network(s) and machine learning (ML) model(s) are implemented for sensor data analysis to perform causality analysis for prognostics. Layer(s) (say last layer) of Deep Network(s) contains the automatically derived features that can be used for ML tasks. Parameter(s) tuning is performed based on the set of features that were recommended by the system to determined performance of systems (or applications) under consideration.

Signal analysis systems and methods for features extraction and interpretation thereof

An application development system for development of Internet of Things (IoT) application includes a cataloging module to obtain an input from an application developer. The input comprises data related to the IoT application to be developed. The cataloging module further retrieves a plurality of reusable artefacts from a knowledge database based on the input. A recommendation module in the application development system recommends, to the application developer, artefacts from amongst the plurality of reusable artefacts, based at least on one of a feedback associated with each of the plurality of reusable artefacts, an expert analysis, and a combination of the expert analysis and the feedback. An association module in the application development system associates artefacts selected by the application developer with each other for development of the IoT application.

Internet of things (IoT) application development

A method and system has been provided for recommending features for developing an IoT analytics application. The method follows a deep like architecture. It comprises of three distinct layers. First layer is for input signal processing and other two layers are for feature reduction. The time domain, frequency domain and time-frequency domain features are extracted from the input signal. The invention uses multiple feature selection methods so that the union of the recommended features by these feature selection methods is significantly lesser than the initial set of features. The best feature combination is recommended using an exhaustive search.

Method and system for recommending features for developing an iot application

A system and method for identifying an unknown person based on a static posture of the unknown person is described. The method includes receiving data of N skeleton joints of the unknown person from a skeleton recording device. The method further includes identifying the static posture of the unknown person. The method includes dividing a skeleton structure of the unknown person in a plurality of body parts based on joint types of the skeleton structure. In addition, the method includes extracting feature vectors for each of the joint type from each of the plurality of body parts. The method further includes identifying the unknown person based on comparison of the feature vectors for the unknown person with one of a constrained feature dataset and an unconstrained feature dataset for a plurality of known persons.

Static posture based person identification

A method for monitoring physiological parameters associated with a subject using a hand held device is described herein. In an implementation, the method includes obtaining a plurality of sample photoplethysmographic (PPG) features associated with a sample subject, from a video of a body part of the sample subject. From among the plurality of sample PPG features, at least one relevant sample PPG feature associated with the physiological parameter, is selected based on a ground truth value of the physiological parameter for the subject. Further, based on the at least one relevant sample PPG feature and the ground truth value of the physiological parameter, a mathematical model indicative of a correlation between the relevant sample PPG feature and the physiological parameter, is determined. The mathematical model can be deployed for monitoring the physiological parameter in real time.

Monitoring physiological parameters

The present disclosure relates to a system and a method for detection of touching characters in a media, characterized by segmentation of adjoining character spaces. In the very first step, an aspect ratio is calculated for each connected component. A candidate touching position of each character is determined by calculating a threshold aspect ratio for each character. Further, a candidate cut column is determined based on a relation between column pixel densities and corresponding length thereof the column in order to segment the touching characters at the candidate cut column.

System and method for detection and segmentation of touching characters for OCR

Disclosed is a method for segmenting a plurality of objects from a two-dimensional (2D) video captured through a depth camera and an RGB/G camera. The method comprises detecting camera motion in each 2D frame of the plurality of 2D frames from the 2D video and generate a first set of 2D frames without any camera motion. The method further comprises generating a plurality of cloud points for the first set of 2D frames corresponding to each pixel associated a 2D frames in the first set of 2D frames. The method further comprises generating a 3D grid comprising a plurality of voxels. The method further comprises determining valid voxels and an invalid voxels in the 3D grid. Further, a 3D connected component labeling technique is applied on to the set of valid voxels to segment the plurality of objects in the 2D video.

Segmenting objects in multimedia data

A system and method for a real-time prognosis of a vehicle comprising a personal communication device comprising an arbitrarily oriented three-axis accelerometer configured to capture a pitch motion and/or roll motion of the vehicle and an onboard diagnostics system communicably connected with the personal communication device enabling bi-directional communication. The personal communication device comprising a processor configured for geometric mapping of a three dimensional Cartesian coordinate of the three-axis accelerometer with the vehicle. The processor virtually orients the coordinates of three-axis accelerometer to coincide with the coordinates of the vehicle. The arbitrarily oriented three-axis accelerometer is configured to capture a road condition and a driver behavior using a sampling rate between 4 Hertz (Hz) to 10 Hertz (Hz). The system for the real-time prognosis of the vehicle, wherein the real-time prognosis utilizes at least one predictive analysis model to determine real-time prognosis for the said vehicle.

System and a method for improved car prognosis

Growing community of inventors

Kolkata, India

Tanushyam Chattopadhyay