PeopleofSuperintelligence

Researcher

Mirco Mutti

Mirco Mutti is a reinforcement-learning researcher whose work studies reward-free exploration and how agents adapt across tasks. He coauthored IDE³AL, a model-based algorithm for reward-free exploration, and his 2022 paper on non-Markovian policies received an ICML Outstanding Paper Award. He completed a 2023 doctoral thesis in Data Science and Computation at the University of Bologna, supervised by Marcello Restelli, and joined the Technion as a postdoctoral researcher in September 2023. His later work examined test-time regret, exploration with parallel agents, and confidence calibration for vision-language-action models. As of October 2026, he is a postdoctoral researcher at the Technion, working with Aviv Tamar in the Robots Learning Lab.

Now at
Technion

Timeline

  1. 2020

    IDE³AL for reward-free exploration

    In 2020, he and Marcello Restelli publish “An Intrinsically-Motivated Approach for Learning Highly Exploring and Fast Mixing Policies.” The paper introduces IDE³AL, a model-based reinforcement-learning algorithm for policies that explore widely and mix quickly.

  2. 2022

    An ICML Outstanding Paper

    In 2022, he, Riccardo De Santi, and Marcello Restelli argue that non-Markovian policies matter for maximum state-entropy exploration in finite samples. ICML gives their paper an Outstanding Paper Award.

  3. 2023

    A doctoral thesis on state entropy

    In 2023, he completes a doctoral thesis in Data Science and Computation at the University of Bologna titled “Unsupervised reinforcement learning via state entropy maximization.” Marcello Restelli supervises the thesis. In September 2023, he joins the Technion as a postdoctoral researcher.

  4. 2024

    Test-time regret in meta-learning

    In 2024, he and Aviv Tamar analyze test-time regret in meta-reinforcement learning. They derive a lower bound that makes the regret rate nearly optimal under a task-separation condition.

  5. 2025

    Exploration with parallel agents

    In 2025, with Vincenzo De Paola, Riccardo Zamboni, and Marcello Restelli, he publishes “Enhancing Diversity In Parallel Agents: A Maximum State Entropy Exploration Story.” The method balances entropy within agents with diversity across agents.

  6. 2026

    Temporal-difference calibration for VLA

    In 2026, with Shelly Francis-Meretzki, Yaniv Romano, and Aviv Tamar, he introduces a sequential Brier-score method for calibration in vision-language-action models. For binary outcomes, its risk minimizer matches the policy’s value function.

  7. As of October 2026

    Postdoctoral researcher at Technion

    As of October 2026, he is a postdoctoral researcher at the Technion, working with Aviv Tamar in the Robots Learning Lab.

Career

  1. Since 2023TechnionPostdoctoral researcher

Education

  1. 2023University of BolognaDoctoral degree in Data Science and Computation

From the biography.

  1. Universities keep AI researchers for a median of 6 years, compared with 2 years at frontier labsIncludes Technion. Across 15,811 dated work stays (6,529 people), median role tenure runs from 6 years at universities to 2 years at frontier labs. We compare all five organization types, starting decades from the 1970s to the 2020s, and the organizations that keep people longest.
  2. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Technion. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. 41% of people who worked abroad later worked in their country of birth among people with a recorded birth country and dated work historyIncludes Technion. Of 991 people with a recorded birth country who worked outside it, 410 (41%) later held a work position at home, led by the United States at 69%.
  4. The United States employs an estimated 48% of people shaping AIIncludes Technion. We estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Mirco Mutti yet. Send us a free-licensed photo.

  1. Vol. 34 No. 04: AAAI-20 Technical Tracks 4 Archives - AAAI aaai.org. Checked 21 October 2026.
  2. [1907.04662] An Intrinsically-Motivated Approach for Learning Highly Exploring and Fast Mixing Policies arxiv.org. Checked 21 October 2026.
  3. The Importance of Non-Markovianity in Maximum State Entropy Exploration proceedings.mlr.press. Checked 21 October 2026.
  4. ICML 2022 Awards icml.cc. Checked 21 October 2026.
  5. Unsupervised reinforcement learning via state entropy maximization amsdottorato.unibo.it. Checked 21 October 2026.
  6. Test-Time Regret Minimization in Meta Reinforcement Learning proceedings.mlr.press. Checked 21 October 2026.
  7. Enhancing Diversity In Parallel Agents: A Maximum State Entropy Exploration Story proceedings.mlr.press. Checked 21 October 2026.
  8. Temporal Difference Calibration in Sequential Tasks: Application to Vision-Language-Action Models arxiv.org. Checked 21 October 2026.
  9. Mirco Mutti muttimirco.github.io. Checked 21 October 2026.
  10. Reinforcement Learning Generalization by Dr Mirco Mutti | College of Computing and Data Science | NTU Singapore ntu.edu.sg. Checked 21 October 2026.