Search this site
Embedded Files
Arpan Losalka

Google Scholar  |  LinkedIn

Hi, I am Arpan, a postdoctoral research fellow in the Department of Computer Science at  the National University of Singapore, working with Prof Jonathan Scarlett.


I primarily conducted research on safety in the context of Bayesian optimization (BO) and bandit algorithms during my PhD, and am currently pursuing research in the area of safe reinforcement learning as a natural extension of the safe BO problem setup.

I have also worked on robustness of both text and image based machine learning models during my tenure at IBM Research AI earlier, and remain interested in topics related to the broader area of trustworthy machine learning and AI safety.



Academic and Work Experience


  • Postdoctoral Research Fellow (Computer Science)   |   National University of Singapore   |   2024 - 2025 (expected)

  • PhD (Computer Science), President's Graduate Fellow   |   National University of Singapore   |   2019 - 2024

  • Research Engineer   |   IBM Research AI, IBM India Research Lab   |   2016 - 2019

  • M.Tech (Computer Science)   |   Indian Statistical Institute Calcutta   |   2014 - 2016

  • Software Engineer   |   Samsung R&D Institute India   |   2013 - 2014

  • B.Tech (Electrical Engineering)   |   IIT Bhubaneswar   |   2009 - 2013


Publications and Patents


  1. No-Regret Algorithms for Safe Bayesian Optimization with Monotonicity Constraints (AISTATS 2024)

Arpan Losalka, Jonathan Scarlett  |  Paper
Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:3232-3240

  1. Benefits of Monotonicity in Safe Exploration with Gaussian Processes (UAI 2023)

Arpan Losalka, Jonathan Scarlett  |  Paper
Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence, PMLR 216:1304-1314

  1. Stochastic linear bandits robust to adversarial attacks (AISTATS 2021)

Ilija Bogunovic, Arpan Losalka, Andreas Krause, Jonathan Scarlett  |  Paper  |  Video

Proceedings of The 24th International Conference on Artificial Intelligence and Statistics, PMLR 130:991-999

  1. Automated testing of dialog systems (US Patent, 2021)

Arpan Losalka, Diptikalyan Saha  |  Patent

US Patent 10,984,198

  1. Resisting adversarial attacks using Gaussian mixture variational autoencoders (AAAI 2019)

Partha Ghosh, Arpan Losalka, Michael J Black  |  Paper  |  Video  |  Blog Post

Proceedings of the AAAI Conference on Artificial Intelligence 33 (01), 541-548

  1. Development of an Automated Examination Seating Arrangement Generation System using Music Inspired Harmony Search Algorithm (SOCPROS 2011)

Arnav Acharyya, Arpan Losalka, Pravir Singh Gupta and Ganapati Panda  |  Paper

Proceedings of the International Conference on Soft Computing for Problem Solving (SocProS 2011) December 20-22, 2011


Projects and Achievements


  • PhD Thesis (NUS)   |   Safety and Reliability in Bayesian Optimization: Algorithms and Regret Bounds (2021-2024)

with Prof Jonathan Scarlett, Department of Computer Science and Mathematics, NUS

  • Research Project (NUS)   |   Stochastic Linear Bandits Robust to Adversarial Attacks (2020-2021)

with Prof Jonathan Scarlett, NUS and Prof Ilija Bogunovich, EEE, University College London

  • Machine Learning Summer School (Tübingen) (2020)

Selected among top 180 applicants for attending the Machine Learning Summer School

  • PhD (NUS)   |   President's Graduate Fellowship Award

for the duration of PhD at NUS

  • PhD (NUS)   |   Teaching Fellowship Award

Received the Teaching Fellowship Award from the NUS School of Computing for teaching excellence 

  • PhD (NUS)   |   Honour List of Student Tutors

Placed on the Honour List of Student Tutors for the Academic Year 2021-22, 2022-23 and 2023-24

  • PhD (NUS)   |   Research Achievement Award, Research Incentive Award

Received the awards in recognition of research contributions (Academic Year 2023-24)

  • IBM Research AI   |   Robustness of text-based AI models (specifically, IBM Watson Assistant) (2017-2019)

with Dr. Diptikalyan Saha, Senior Technical Staff Member, IBM Research AI

  • IBM Research AI   |   Manager's Choice Awards (2018-2019)

Received the award twice during my tenure at IBM Research AI

  • IBM Research   |   Biologically Inspired Information Processing in Machine Learning (2016-2017)

with Dr. Ravi Kothari, then Chief Scientist, IBM India Research Lab

  • M.Tech Dissertation (ISI Calcutta)   |   On the Classification of Vertically Partitioned Data, the Naive Bayes Classifier and the Independence Assumption (2015-2016)

with Prof C.A. Murthy, Machine Intelligence Unit, ISI Calcutta and Dr. Ravi Kothari, then Chief Scientist, IBM India Research Lab

  • M.Tech (ISI Calcutta)   |   Semester Prize for Academic Excellence (2014-2016)

Received the award in all semesters at ISI Calcutta

  • Summer Internship (IBM Research)   |   Insights into Streaming Data using Apache Spark (2015)

with Dr. Ravi Kothari, then Chief Scientist, IBM India Research Lab

  • B.Tech Dissertation (IIT Bhubaneswar)   |   Development of an Automated Music Transcription System (2013)

with Prof. Ganapati Panda, then Head of School, School of Electrical Sciences and Deputy Director, IIT Bhubaneswar

  • Summer Internship (UMASS Dartmouth)   |   Characterizing the Properties of a Sound Attenuating Room and Learning How to Make Recordings (2012)

with Prof. Karen Payton, now retired, UMass Dartmouth Professor Emerita

  • Others

ISI Entrance Exam (B.Math): selected among top 128 nationally (2009)

CMI Entrace Exam (B.Sc): selected among top 50 nationally (2009)

Secondary Exam (WBBSE): ranked 4th at state level in West Bengal, received Mamraj Agarwal National Award, The Telegraph School Award for Excellence, invited to participate in in the CSIR Programme on Youth for Leadership in Science (2007)



 

Please reach out to me at arpan [at] nus {dot} followed by edu <dot> sg

 

Google Sites
Report abuse
Page details
Page updated
Google Sites
Report abuse