Vijay Arya

IBM Research - India

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Thierry Turletti

Institut national de recherche en informatique et en automatique

Vikas Chandan

Pacific Northwest National Laboratory

Shivkumar Kalyanaraman

Microsoft (Finland)

Kumar Saurav

Tata Institute of Fundamental Research

Amit Dhurandhar

IBM (United States)

Karthikeyan Shanmugam

PSG Institute of Medical Sciences & Research

Sameep Mehta

IBM Research - India

Sambaran Bandyopadhyay

IBM Research - India

Deva P. Seetharam

IBM Research - India

Sami Mourad

IBM (United States)

Cooperative Institutions

IBM (United States)

IBM Research - India

IBM Research - Thomas J. Watson Research Center

Wyckoff Heights Medical Center

IBM (India)

Indian Institute of Technology Bombay

Universiti Brunei Darussalam

Indian Institute of Technology Delhi

Alliance for Safe Kids

University of Massachusetts Amherst

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

AI Explainability 360: An Extensible Toolkit for Understanding Data and Machine Learning Models

Journal of Machine Learning Research (2020)

Vijay Arya Rachel Bellamy Pin‐Yu Chen Amit Dhurandhar Michael Hind

Extensibility

Source

Cite

Citations (56)

Global Explanations for Multivariate time series models

Vijay Arya Diptikalyan Saha Sandeep Hans Amaresh Rajasekharan Tony Tang

Several explainable AI algorithms have been proposed to help make machine learning models more interpretable and trustworthy. However in spite of numerous methodological advancements, there is still a persistent gap between what researchers develop and what business users seek. In this work, we aim to bridge this gap for an AI system that predicts the remaining useful life of an aircraft's engine using time series data collected from multiple sensors. We propose a novel approach to compute easily understandable explanations by fusing two explainers in sequence wherein explanations of the first explainer are explained by the second. We use this approach to build a global post-hoc model-agnostic explainer for AI models that ingest multivariate time series data. Our approach fuses a local explainer that yields feature importance weights, with a directly interpretable model that outputs global rules. Our experimental results based on the C-MAPSS open-source dataset demonstrate that the proposed two-stage explainer computes global explanations that are amenable to business users and sheds light on how the behavior of an individual and a group of sensors impacts the remaining useful life of an aircraft's engine.

10.1145/3570991.3570998

Cite

Citations (2)

Complete Resolution of Anemia after Polypectomy of Inflammatory Fibroid Polyp

The American Journal of Gastroenterology (2007)

Vijay Arya Siddharth Mathur Niket Sonpal

Purpose: Inflammatory fibroid polyp is an uncommon benign polypoid lesion of the gastrointestinal tract. Inflammatory fibroid polyp is more commonly found in the stomach or small intestine, and rarely in the colon. We present a rare case of an 85 year old female (Jehovah's Witness), presenting with iron deficiency anemia caused by a benign fibroid polyp in the colon. Methods: An 85 year old Hispanic female with a past medical history of HTN, vaginal hysterectomy and family history of colon cancer, was referred to our hospital (a center for bloodless surgery), after being found to have a large colonic polyp in ascending colon and iron deficiency anemia. Her Hg was 10.1, HCT 31.5, MCV 78.2, and her stool for occult blood was positive. On colonoscopy, diverticulosis was seen from descending to sigmoid colon and in the ascending colon a large pedunculated polyp with a very thick stalk was found. At the stalk site 6 cc epinephrine, 1 in 10,000 was injected and the polyp was removed using snare polypectomy. Pathology reported the mass as an inflammatory fibroid polyp with multiple erosions measuring 4 × 3 × 3 cm in size. An EGD showed mild hiatal hernia, gastric erosions and mild duodenitis. Patient was never given any iron supplements or erythropoietin injections for the anemia. Follow up labs after 10 months demonstrated normal levels of Hb of 14.6 and Hct 41.7. Thus this patient had an iron deficiency anemia secondary to blood loss from the fibroid polyp which had multiple areas of erosions on its surface, and the anemia resolved on removal of the polyp. Conclusion: Inflammatory fibroid polyp is histologically characterized by a mixture of proliferating fibroblasts and small blood vessels, accompanying a marked eosinophilic infiltrate. The lesion largely affects adults and is more common in the antrum of the stomach, but has occasionally been reported in the small bowel and colon. Although it is generally believed to represent a reactive, nonneoplastic condition, their histogenesis remains unclear. The treatment is surgical excision of the polyp, or colonoscopic resection when it is possible.

Inflammatory fibroid polyp

Ascending colon

Diverticulosis

Descending colon

Polypectomy

Hiatal Hernia

10.14309/00000434-200709002-00408

Cite

Citations (0)

Recovery of a Sparse Integer Solution to an Underdetermined System of Linear Equations

arXiv (Cornell University) (2011)

T. S. Jayram Soumitra Pal Vijay Arya

We consider a system of m linear equations in n variables Ax=b where A is a given m x n matrix and b is a given m-vector known to be equal to Ax' for some unknown solution x' that is integer and k-sparse: x' in {0,1}^n and exactly k entries of x' are 1. We give necessary and sufficient conditions for recovering the solution x exactly using an LP relaxation that minimizes l1 norm of x. When A is drawn from a distribution that has exchangeable columns, we show an interesting connection between the recovery probability and a well known problem in geometry, namely the k-set problem. To the best of our knowledge, this connection appears to be new in the compressive sensing literature. We empirically show that for large n if the elements of A are drawn i.i.d. from the normal distribution then the performance of the recovery LP exhibits a phase transition, i.e., for each k there exists a value m' of m such that the recovery always succeeds if m > m' and always fails if m < m'. Using the empirical data we conjecture that m' = nH(k/n)/2 where H(x) = -(x)log_2(x) - (1-x)log_2(1-x) is the binary entropy function.

Underdetermined system

Matrix (chemical analysis)

Linear equation

Matrix norm

Source

Cite

Citations (2)

Software-only occupancy inference in a workplace findings from a field trial

Mohit Jain Vikas Chandan Ashok Kumar Vijay Arya R. Sridhar

HVAC and lighting loads contribute a significant fraction of total energy consumed in office buildings. These loads vary as a function of occupancy and therefore inferring occupancy is vital to optimizing energy efficiency within these buildings. This work presents evaluation and comparison results from a field trial conducted in a large office building, which involved estimating occupancy with the help of existing opportunistic context sources versus instrumented hardware sensors. Our results show that opportunistic sensing yielded an accuracy of 80% in comparison with expensive hardware sensors and may be used to continuously estimate fine-grained workplace occupancy in an inexpensive manner. Moreover the inferred occupancy information may also be used to identify anomalies in thermal management and space utilization within the building.

Occupancy

Field trial

Post-occupancy evaluation

10.1109/isgt.2016.7781282

Cite

Citations (1)

Drugs as an Etiology of Acute Pancreatitis in the Era of Polypharmacy in a Teaching Community Hospital

The American Journal of Gastroenterology (2012)

Hemant Goyal Ashok Valluri Samantha Daniel Zainab Bhojani Vijay Arya

Purpose: Drugs have generally been considered to be a relatively uncommon cause of acute pancreatitis (AP), with an estimated incidence of 0.1-2%. However, a recent report of 170 cases of acute pancreatitis at a single academic medical center in Czech Republic concluded that drugs were the most likely cause in 5.3% of cases of AP, making drugs the third most frequent cause of AP after gallstones and alcohol. Another Dutch observational study [EARL study] reported that almost 12.5% of the patients with AP used pancreatitis-associated drug in the absence of another etiologic cause. We have a significant number of patients admitted with diagnosis of AP and recurrent pancreatitis where etiology of pancreatitis was not found. It is possible some of those patients could have pancreatitis related to drug use. Aim: To review the charts of those patients retrospectively and to find out if any patients were taking drugs which could have precipitated the episode of acute pancreatitis. Methods: An IRB approved retrospective study was performed on patients who were admitted in WHMC with primary diagnosis of AP from May 2010 to December 2011, to look for the etiology of pancreatitis. Classification system of drug-induced acute pancreatitis was used according to Badalov et al. Exclusion criteria: patients with history of 1) alcoholism, 2) gallstones, 3) hypertriglyceridemia, 4) post-ERCP 5) hypercalcemia, and 6) pancreatic mass/cyst. Results: Twenty-eight patients were found to be eligible for inclusion in the study. Twenty-one patients were found to be taking at least one drug implicated as cause of drug-induced pancreatitis (DIP) (p<0.01). Only in two patients with AP, the cause of pancreatitis was suspected due to drugs. Four patients were found to have two or more admissions in the study period, and these patients were on more than one drug implicated for pancreatitis. Metformin was the most prescribed drug in (total eight) followed by Simvastatin (total six). Conclusion: This study highlights the importance of medications as an etiology of AP. DIP was found to be more common than diagnosed and often overlooked. Prevention and treatment of DIP need updated knowledge of drugs associated with it and physicians should be more vigilant about the possibility of DIP.

Etiology

10.14309/00000434-201210001-00216

Cite

Citations (0)

Evaluating demand response programs by means of key performance indicators

George Thanos Marilena Minou Tanuja Ganu Vijay Arya Dipanjan Chakraborty

Demand response (DR) has received significant attention in recent years and several DR programs are being deployed and evaluated worldwide. DR systems provide a wide range of economic and operational benefits to different stakeholders of the electrical power system including consumers, generators and distributors. DR can be achieved through a number of different mechanisms such as direct-load-control, incentives, pricing signals, or a combination of these schemes. Due to the remarkable variation in demand response systems, it becomes a challenge to evaluate and compare the effectiveness of different DR programs holistically. In this work, we define a number of different performance metrics that could be used to evaluate DR programs based on peak reduction, demand variation and reshaping, and economic benefits.

Demand Response

Load management

Peak demand

Power demand

10.1109/comsnets.2013.6465597

Cite

Citations (14)

LIPASE/AMYLASE RATIO

The American Journal of Gastroenterology (2004)

Vijay Arya Reza Rezai Aperna Reddy Manjinder Singh Beata Malinowski

Purpose: To asses the ability of lipase/amylase ratio to establish the etiology of pancreatitis. Methods: Charts from 159 patients with aadmitting diagnosis of pancreatitis were reviewed. Three groups were established (table 1). We gave a general heading forpatients with biliary pancreatitis to include patients with: 1) pancreatitis post ERCP 2) pancreatitis secondary to a common bile duct stone/obstruction, 3) post cholecystectomy syndrome (n = 46). Nonbiliary, nonalcoholic (NBNA) patients included patients with pancreatitis secondary to drugs, ischemia, infection, hypertriglyceremia and pancreatic adenocarcinoma.Table 1: Summery of ResultsResults: A considerable overlap was observed between the 3 groups. No statistically significant differences were found between NBNA patients and those with either biliary or alcoholic forms of the disease. The serum lipase/amylase ratios in patients with alcoholic pancreatitis ranged from [.014 to 1.7], in those with biliary pancreatitis from [.049-1.48], and in those with NBNA pancreatitis from [.041–.83] These differences were not statistically significant. On admission amylase, was significantly lower in alcohol induced pancreatitis than in patients with biliary pancreatitis Conclusions: Even though amylase, was significantly lower in alcoholics than in patients with biliary pancreatitis, there was a wide range in the L/A ratio in all 3 groups and comparison of the median value between these groups were not statistically significant. Prospective studies in literature have shown that the L/A ratio greater than 2/1 are more specific for alcohol induced pancreatitis. In our study we observed the highest L/A to be [1.7]. The lipase to amylase ratio does not appear to be sufficiently sensitive or specific to distinguish alcoholic from nonalcoholic pancreatitis. We conclude that a prospective study with a larger number of patients is needed to re-evaluate the clinical efficacy of this ratio.

Etiology

Pancreatic Disease

10.14309/00000434-200410001-00170

Cite

Citations (0)

Explicit routing in multicast overlay networks

Computer Communications (2006)

Torsten Braun Vijay Arya Thierry Turletti

IP multicast

Tree (set theory)

Overlay multicast

10.1016/j.comcom.2006.02.022

Cite

Citations (5)

Blockchain Enabled Trustless API Marketplace

Vijay Arya Sayandeep Sen Palani Kodeswaran

There has been an unprecedented surge in the number of service providers offering a wide range of machine learning prediction APIs for tasks such as image classification, language translation, etc. thereby monetizing the underlying data and trained models. Typically, a data owner (API provider) develops a model, often over proprietary data, and leverages the infrastructure services of a cloud vendor for hosting and serving API requests. Clearly, this model assumes complete trust between the API Provider and cloud vendor. On the other hand, a malicious/buggy cloud vendor may copy the APIs and offer an identical service, under-report model usage metrics, or unfairly discriminate between different API providers by offering them a nominal share of the revenue. In this work, we present the design of a blockchain based decentralized trustless API marketplace that enables all the stakeholders in the API ecosystem to audit the behavior of the parties without having to trust a single centralized entity. In particular, our system divides an AI model into multiple pieces and deploys them among multiple cloud vendors who then collaboratively execute the APIs. Our design ensures that cloud vendors cannot collude with each other to steal the combined model, while individual cloud vendors and clients cannot repudiate their input or model executions.

Cloud service provider

10.1109/comsnets48256.2020.9027456

Cite

Citations (6)