Lars Lien Ankile

ML Research @ MIT


44 Vassar St

Cambridge, MA 02139

I’m an M.Eng. of Data Science student at Harvard University, currently working on my thesis at MIT CSAIL in Prof. Pulkit Agrawal’s Improbable AI group, where I’m currently investigating learning-based methods for robotic manipulation.

Before this, I spent a year in the Data to Actionable Knowledge Lab at Harvard, working with Profs. Weiwei Pan and Finale Doshi-Velez on applying RL and Bayesian inference to model human decision-making for frictionful tasks in healthcare settings. I also worked with Prof. David Parkes and Matheus Ferreira in the EconCS Lab at Harvard working on detecting manipulation in multi-agent settings.

I did my undergrad at the Norwegian University of Science and Technology (NTNU) and did my thesis work on applying Deep RL to econometric forecasting of complex and multivariate time series, supervised by Prof. Sjur Westgaard.

Research Interests

My research goal is to enable machines to learn in a human-like manner, primarily through observation and interaction by developing methods merging imitation and reinforcement learning with robust policy representations. I aim to bridge the current divide between general policies that address simple tasks and more specialized policies tailored for complex tasks.

Here’s a sneek peek into our ongoing project on making capable behavior model for furniture assembly tasks:

Furniture Assembly


Apr 10, 2024 Our recent work on Data-Efficient Imitation Learning for Robotic Assembly is available to read on arXiv! In this work, we show how one can learn long assemblies (~2500 timesteps) with only <50 demonstrations using diffusion policies and data augmentation strategies.

selected publications

  1. arXiv
    JUICER: Data-Efficient Imitation Learning for Robotic Assembly
    Lars Ankile, Anthony Simeonov, Idan Shenfeld, and 1 more author
  2. I See You! Robust Measurement of Adversarial Behavior
    Lars Ankile, Matheus XV Ferreira, and David Parkes
    In Multi-Agent Security Workshop @ NeurIPS’23 2023
  3. Discovering User Types: Mapping User Traits by Task-Specific Behaviors in Reinforcement Learning
    Lars Lien Ankile, Brian Ham, Kevin Mao, and 4 more authors
    In First Workshop on Theory of Mind in Communicating Agents @ ICML’23 2023
  4. M.Sc.
    Exploration of Forecasting Paradigms and a Generalized Forecasting Framework
    Lars Lien Ankile, and Kjartan Krange