公开(公告)号：US20200285991A1

公开(公告)日：2020-09-10

申请号：US16542339

申请日：2019-08-16

Applicant: Tata Consultancy Services Limited

Inventor： Satanik PANDA ,  Abhishek ROY CHOUDHURY ,  Prateep MISRA ,  Tanushyam CHATTOPADHYAY

IPC: G06N20/00 ,  G06F21/62 ,  H04L9/06

Abstract: 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.

公开(公告)号：US20190279090A1

公开(公告)日：2019-09-12

申请号：US16265906

申请日：2019-02-01

Applicant: Tata Consultancy Services Limited

Inventor： Tulika BOSE ,  Angshul MAJUMDAR ,  Tanushyam CHATTOPADHYAY

IPC: G06N3/08 ,  G06N3/04

Abstract: 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.

公开(公告)号：US20160125253A1

公开(公告)日：2016-05-05

申请号：US14889946

申请日：2014-04-25

Applicant: Tata Consultancy Services Limited

Inventor： Aishwarya VISVANATHAN ,  Tanushyam CHATTOPADHYAY

IPC: G06K9/32 ,  G06K9/62 ,  G06K9/40 ,  G06K9/46

CPC classification number: G06K9/3258 ,  G06K9/40 ,  G06K9/4671 ,  G06K9/6202 ,  G06K9/6211 ,  G06K2209/01 ,  G06K2209/15

Abstract: Disclosed is a method and apparatus for computation and processing of an image for image matching. The apparatus here is configured to pre-process plurality of images for creating an image template. Next, the test image is extracted and pre-processed for assessing the degree of match between the test image components and the image components of the images in the image template, based a position based matching score, a feature based matching score or both.

Abstract translation: 公开了一种用于图像匹配的图像的计算和处理的方法和装置。 这里的装置被配置为预处理用于创建图像模板的多个图像。 接下来，基于基于位置的匹配分数，基于特征的匹配分数或两者来提取测试图像并对其进行预处理，以评估测试图像分量与图像模板中的图像的图像分量之间的匹配度。

公开(公告)号：US20140140622A1

公开(公告)日：2014-05-22

申请号：US14081871

申请日：2013-11-15

Applicant: TATA CONSULTANCY SERVICES LIMITED

Inventor： Tanushyam CHATTOPADHYAY ,  Bidyut B. CHAUDHURI

IPC: G06K9/18

CPC classification number: G06K9/18 ,  G06K9/3266 ,  G06K9/6282

Abstract: The present disclosure relates to designing of a hierarchy of feature vectors. In one embodiment, a method for facilitating design of a hierarchy of feature vectors while recognizing one or more characters in a video is disclosed. The method comprises collecting one or more features from each of the segments in a video frame extracted from a video; preparing multi-dimensional feature vectors to classify the one or more characters; calculating a minimum distance between the multi-dimensional features vectors of a test character and the multi-dimensional feature vectors of a pre-stored character template; selecting, with respect to a decreasing order of the minimum distance, the multi-dimensional feature vectors to design a hierarchy of the multi-dimensional feature vectors; and classifying the characters based on the hierarchy of the multi-dimensional feature vectors.

Abstract translation: 本公开涉及特征向量的层次结构的设计。 在一个实施例中，公开了一种便于在识别视频中的一个或多个字符的同时设计特征向量的层次的方法。 该方法包括从从视频提取的视频帧中的每个段收集一个或多个特征; 准备多维特征向量来分类一个或多个字符; 计算测试人物的多维特征向量与预先存储的角色模板的多维特征向量之间的最小距离; 针对所述最小距离的递减顺序选择所述多维特征向量来设计所述多维特征向量的层级; 并且基于多维特征向量的层次来对角色进行分类。