Pratik Gajane

@tue.nl

postdoctoral researcher, Mathematics and Computer Science
Eindhoven University of Technology (

Pratik Gajane


                     Download Compact PDF  
https://researchid.co/pratik.gajane

RESEARCH INTERESTS

My research interests span across sequential decision making and fairness in machine learning.
13

Scopus Publications

1000

Scholar Citations

12

Scholar h-index

12

Scholar i10-index

Scopus Publications

  • Multi-armed Bandits with Generalized Temporally-Partitioned Rewards
    Ronald C. van den Broek, Rik Litjens, Tobias Sagis, Nina Verbeeke, Pratik Gajane
    Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics, 2024
  • WeHeart: A Personalized Recommendation Device for Physical Activity Encouragement in Cardiac Rehabilitation
    Rosa Van Tuijn, Tianqin Lu, Emma Driesse, Koen Franken, Pratik Gajane, et al.
    Frontiers in Artificial Intelligence and Applications, 2023
    We introduce WeHeart, a personalized recommendation device that aims to gradually increase physical activity levels in cardiac rehabilitation. The importance of physical activity in cardiac rehabilitation as a means of reducing associated morbidity and mortality rates is well-established. However, forming physical activity habits is a challenge, and the approach varies depending on individual preferences. Our solution employs a Random Forest classification model that combines both measured and self-reported data to provide personalized recommendations. We also propose to make use of Explainable AI to improve transparency and foster trust.
  • Autonomous Exploration for Navigating in MDPs Using Blackbox RL Algorithms
    Pratik Gajane, Peter Auer, Ronald Ortner
    Ijcai International Joint Conference on Artificial Intelligence, 2023
    We consider the problem of navigating in a Markov decision process where extrinsic rewards are either absent or ignored. In this setting, the objective is to learn policies to reach all the states that are reachable within a given number of steps (in expectation) from a starting state. We introduce a novel meta-algorithm which can use any online reinforcement learning algorithm (with appropriate regret guarantees) as a black-box. Our algorithm demonstrates a method for transforming the output of online algorithms to a batch setting. We prove an upper bound on the sample complexity of our algorithm in terms of the regret bound of the used black-box RL algorithm. Furthermore, we provide experimental results to validate the effectiveness of our algorithm and correctness of our theoretical results.
  • LEMON: Alternative Sampling for More Faithful Explanation Through Local Surrogate Models
    Dennis Collaris, Pratik Gajane, Joost Jorritsma, Jarke J. van Wijk, Mykola Pechenizkiy
    Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics, 2023
    Local surrogate learning is a popular and successful method for machine learning explanation. It uses synthetic transfer data to approximate a complex reference model. The sampling technique used for this transfer data has a significant impact on the provided explanation, but remains relatively unexplored in literature. In this work, we explore alternative sampling techniques in pursuit of more faithful and robust explanations, and present LEMON: a sampling technique that samples directly from the desired distribution instead of reweighting samples as done in other explanation techniques (e.g., LIME). Next, we evaluate our technique in a synthetic and UCI dataset-based experiment, and show that our sampling technique yields more faithful explanations compared to current state-of-the-art explainers.
  • WeHeart: A Personalized Recommendation Device for Physical Activity Encouragement and Preventing “Cold Start” in Cardiac Rehabilitation
    Rosa van Tuijn, Tianqin Lu, Emma Driesse, Koen Franken, Pratik Gajane, et al.
    Lecture Notes in Computer Science, 2023
  • The Impact of Batch Learning in Stochastic Linear Bandits
    Danil Provodin, Pratik Gajane, Mykola Pechenizkiy, Maurits Kaptein
    Proceedings IEEE International Conference on Data Mining Icdm, 2022
    We consider a special case of bandit problems, named batched bandits, in which an agent observes batches of responses over a certain time period. Unlike previous work, we consider a more practically relevant batch-centric scenario of batch learning. That is to say, we provide a policy-agnostic regret analysis and demonstrate upper and lower bounds for the regret of a candidate policy. Our main theoretical results show that the impact of batch learning is a multiplicative factor of batch size relative to the regret of online behavior. Primarily, we study two settings of the stochastic linear bandits: bandits with finitely and infinitely many arms. While the regret bounds are the same for both settings, the former setting results hold under milder assumptions. Also, we provide a more robust result for the 2-armed bandit problem as an important insight. Finally, we demonstrate the consistency of theoretical results by conducting empirical experiments and reflect on optimal batch size choice.
  • Gambler bandits and the regret of being ruined
    Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems Aamas, 2021
  • Variational regret bounds for reinforcement learning
    35th Conference on Uncertainty in Artificial Intelligence Uai 2019, 2019
  • Adaptively Tracking the Best Bandit Arm with an Unknown Number of Distribution Changes
    Proceedings of Machine Learning Research, 2019
  • Variational Regret Bounds for Reinforcement Learning
    Proceedings of Machine Learning Research, 2019
  • Achieving Optimal Dynamic Regret for Non-stationary Bandits without Prior Information
    Proceedings of Machine Learning Research, 2019
  • Corrupt Bandits for Preserving Local Privacy
    Proceedings of Machine Learning Research, 2018
  • A relative exponential weighing algorithm for adversarial utility-based dueling bandits
    32nd International Conference on Machine Learning Icml 2015, 2015

RECENT SCHOLAR PUBLICATIONS

  • Best-of-Both-Worlds Multi-Dueling Bandits: Unified Algorithms for Stochastic and Adversarial Preferences under Condorcet and Borda Objectives
    S Akash, P Gajane, J Singh
    arXiv preprint arXiv:2603.18972 , 2026
    2026
  • Evaluating Causal Discovery Algorithms for Path-Specific Fairness and Utility in Healthcare
    N Nagesh, E Khatibi, T Hughes, M Bagheri, P Gajane, AM Rahmani
    arXiv preprint arXiv:2603.15926 , 2026
    2026
  • Adversarial Multi-dueling Bandits
    P Gajane
    arXiv preprint arXiv:2406.12475 , 2024
    2024
    Citations: 2
  • Multi-Armed Bandits with Generalized Temporally-Partitioned Rewards.
    RC van den Broek, R Litjens, T Sagis, L Siecker, N Verbeeke, P Gajane
    22nd Symposium on Intelligent Data Analysis (IDA) , 2024
    2024
  • Investigating Gender Fairness in Machine Learning-driven Personalized Care for Chronic Pain
    P Gajane, S Newman, JD Piette
    https://arxiv.org/abs/2402.19226 , 2024
    2024
    Citations: 4
  • Provably Efficient Exploration in Constrained Reinforcement Learning: Posterior Sampling Is All You Need
    D Provodin, P Gajane, M Pechenizkiy, M Kaptein
    arXiv preprint arXiv:2309.15737 , 2023
    2023
  • University Teaching Qualification Basiskwalificatie Onderwijs (BKO) Teaching Portfolio
    P Gajane
    2023
  • Autonomous Exploration for Navigating in MDPs Using Blackbox RL Algorithms.
    P Gajane, P Auer, R Ortner
    IJCAI, 3714-3722 , 2023
    2023
    Citations: 1
  • WeHeart: A Personalized Recommendation Device for Physical Activity Encouragement and Preventing “Cold Start” in Cardiac Rehabilitation
    PGEB Rosa van Tuijn, Tianqin Lu, Emma Driesse, Koen Franken
    Human-Computer Interaction – INTERACT 2023. 14144 (Lecture Notes in Computer … , 2023
    2023
    Citations: 3
  • LEMON: Alternative sampling for more faithful explanation through local surrogate models
    D Collaris, P Gajane, J Jorritsma, JJ van Wijk, M Pechenizkiy
    International Symposium on Intelligent Data Analysis, 77-90 , 2023
    2023
    Citations: 13
  • Multi-Armed Bandits with Generalized Temporally-Partitioned Rewards
    RC van den Broek, R Litjens, T Sagis, L Siecker, N Verbeeke, P Gajane
    16th European Workshop on Reinforcement Learning (EWRL), arXiv: 2303.00620 , 2023
    2023
    Citations: 1
  • Curiosity-driven Exploration in Sparse-reward Multi-agent Reinforcement Learning
    J Li, P Gajane
    16th European Workshop on Reinforcement Learning (EWRL) , 2023
    2023
    Citations: 15
  • Local Differential Privacy for Sequential Decision Making in a Changing Environment
    P Gajane
    Fourth AAAI Workshop on Privacy-Preserving Artificial Intelligence , 2023
    2023
    Citations: 1
  • Industrializing Deep Reinforcement Learning for ASML’s Service Network
    JFJ van der Haar, IRJIR Basten, WW van Jaarsveld, PP Gajane, ...
    Master’s thesis, Eindhoven University of Technology, Eindhoven, The Netherlands , 2023
    2023
    Citations: 1
  • The impact of batch learning in stochastic linear bandits
    D Provodin, P Gajane, M Pechenizkiy, M Kaptein
    2022 IEEE International Conference on Data Mining (ICDM), 1149-1154 , 2022
    2022
    Citations: 6
  • Generalizing distribution of partial rewards for multi-armed bandits with temporally-partitioned rewards
    RC Broek, R Litjens, T Sagis, L Siecker, N Verbeeke, P Gajane
    arXiv preprint arXiv:2211.06883 , 2022
    2022
  • An Empirical Evaluation of Posterior Sampling for Constrained Reinforcement Learning
    D Provodin, P Gajane, M Pechenizkiy, M Kaptein
    Reinforcement Learning for Real Life Workshop , 2022
    2022
  • Survey on fair reinforcement learning: Theory and practice
    P Gajane, A Saxena, M Tavakol, G Fletcher, M Pechenizkiy
    arXiv preprint arXiv:2205.10032 , 2022
    2022
    Citations: 38
  • The impact of batch learning in stochastic bandits
    D Provodin, P Gajane, M Pechenizkiy, M Kaptein
    Workshop on Ecological Theory of Reinforcement Learning , 2021
    2021
    Citations: 2
  • Gambler bandits and the regret of being ruined
    FS Perotto, S Vakili, P Gajane, Y Faghan, M Bourgais
    20th International Conference on Autonomous Agents and Multiagent Systems … , 2021
    2021
    Citations: 7

MOST CITED SCHOLAR PUBLICATIONS

  • On formalizing fairness in prediction with machine learning
    P Gajane, M Pechenizkiy
    the 5th Workshop on Fairness, Accountability, and Transparency in Machine … , 2018
    2018
    Citations: 350
  • Adaptively tracking the best bandit arm with an unknown number of distribution changes
    P Auer, P Gajane, R Ortner
    Conference on learning theory, 138-158 , 2019
    2019
    Citations: 185
  • Variational regret bounds for reinforcement learning
    R Ortner, P Gajane, P Auer
    Uncertainty in Artificial Intelligence, 81-90 , 2020
    2020
    Citations: 80
  • A sliding-window algorithm for markov decision processes with arbitrarily changing rewards and transitions
    P Gajane, R Ortner, P Auer
    Lifelong Learning: A Reinforcement Learning Approach Workshop at FAIM , 2018
    2018
    Citations: 69
  • A relative exponential weighing algorithm for adversarial utility-based dueling bandits
    P Gajane, T Urvoy, F Clérot
    International Conference on Machine Learning, 218-227 , 2015
    2015
    Citations: 64
  • Corrupt bandits for preserving local privacy
    P Gajane, T Urvoy, E Kaufmann
    Algorithmic Learning Theory, 387-412 , 2018
    2018
    Citations: 47
  • Achieving optimal dynamic regret for non-stationary bandits without prior information
    P Auer, Y Chen, P Gajane, CW Lee, H Luo, R Ortner, CY Wei
    Conference on Learning Theory, 159-163 , 2019
    2019
    Citations: 39
  • Survey on fair reinforcement learning: Theory and practice
    P Gajane, A Saxena, M Tavakol, G Fletcher, M Pechenizkiy
    arXiv preprint arXiv:2205.10032 , 2022
    2022
    Citations: 38
  • Adaptively tracking the best arm with an unknown number of distribution changes
    P Auer, P Gajane, R Ortner
    European Workshop on Reinforcement Learning 14, 375 , 2018
    2018
    Citations: 34
  • Corrupt bandits
    P Gajane, T Urvoy, E Kaufmann
    EWRL , 2016
    2016
    Citations: 17
  • Curiosity-driven Exploration in Sparse-reward Multi-agent Reinforcement Learning
    J Li, P Gajane
    16th European Workshop on Reinforcement Learning (EWRL) , 2023
    2023
    Citations: 15
  • LEMON: Alternative sampling for more faithful explanation through local surrogate models
    D Collaris, P Gajane, J Jorritsma, JJ van Wijk, M Pechenizkiy
    International Symposium on Intelligent Data Analysis, 77-90 , 2023
    2023
    Citations: 13
  • Utility-based dueling bandits as a partial monitoring game
    P Gajane, T Urvoy
    In the 12th European Workshop on Reinforcement Learning (EWRL), 2015 , 2015
    2015
    Citations: 8
  • Gambler bandits and the regret of being ruined
    FS Perotto, S Vakili, P Gajane, Y Faghan, M Bourgais
    20th International Conference on Autonomous Agents and Multiagent Systems … , 2021
    2021
    Citations: 7
  • The impact of batch learning in stochastic linear bandits
    D Provodin, P Gajane, M Pechenizkiy, M Kaptein
    2022 IEEE International Conference on Data Mining (ICDM), 1149-1154 , 2022
    2022
    Citations: 6
  • Investigating Gender Fairness in Machine Learning-driven Personalized Care for Chronic Pain
    P Gajane, S Newman, JD Piette
    https://arxiv.org/abs/2402.19226 , 2024
    2024
    Citations: 4
  • Autonomous exploration for navigating in non-stationary CMPs
    P Gajane, R Ortner, P Auer, C Szepesvari
    arXiv preprint arXiv:1910.08446 , 2019
    2019
    Citations: 4
  • Counterfactual learning for machine translation: Degeneracies and solutions
    C Lawrence, P Gajane, S Riezler
    arXiv preprint arXiv:1711.08621 , 2017
    2017
    Citations: 4
  • WeHeart: A Personalized Recommendation Device for Physical Activity Encouragement and Preventing “Cold Start” in Cardiac Rehabilitation
    PGEB Rosa van Tuijn, Tianqin Lu, Emma Driesse, Koen Franken
    Human-Computer Interaction – INTERACT 2023. 14144 (Lecture Notes in Computer … , 2023
    2023
    Citations: 3
  • Corrupt bandits for privacy preserving input
    P Gajane, T Urvoy, E Kaufmann
    arXiv preprint arXiv:1708.05033 , 2017
    2017
    Citations: 3