I'm a PhD student at MIT CSAIL in the Learning and Intelligence Systems working in the intersection of Machine Learning and Robotics with Leslie Kaelbling and Tomás Lozano-Pérez. Funded by a "La Caixa" fellowship.

Previously, I studied two degrees (Mathematics and Engineering Physics) at UPC in Barcelona thanks to CFIS: an intensive program that selects 40 students from around Spain to study two degrees in 4.5 years. Within CFIS, I was the valedictorian of my promotion.

Papers

Neural Relational Inference with Fast Modular Meta-learning; Preprint

Alet, Weng, Lozano-Perez, Kaelbling; Full paper

Graph Element Networks: structured, adaptive learning and memory; ICML '19

[Long talk: 20% of accepted papers, 4.5% of all submissions]

Alet, Jeewajee, Bauza, Rodriguez, Lozano-Perez, Kaelbling; ICML talk (16min) ; Full paper ; Github code ; ICML slides

Modular meta-learning; CoRL 2018

Alet, Lozano-Perez, Kaelbling; video summary Full paper Github code

Modular meta-learning in abstract graph networks for combinatorial generalization; NeurIPS'18 meta-learning workshop

Alet, Bauza, Rodriguez, Lozano-Perez, Kaelbling; Full paper Github code

OmniPush dataset: accurate, diverse, real-world dataset of pushing dynamics; NeurIPS'18 physics workshop

Bauza*, Alet*, Lozano-Perez, Kaelbling, Rodriguez; Website Full paper Download the data

Finding frequent entities in continuous data; IJCAI '18

Alet, Chitnis, Kaelbling, Lozano-Perez; Slides for IJCAI, Link to arxiv , project website

Robotic Pick-and-Place of Novel Objects in Clutter with Multi-Affordance Grasping and Cross-Domain Image Matching; Zeng et. al. ; ICRA '18

Zeng, Song, Yu et. al. Project website Link to arxiv

Past projects

Amazon Robotics Challege

In 2016 I was a visiting student in the MCube lab at MIT with professor Alberto Rodriguez. There, I was responsible for the High Level planner of team MIT-Princeton in the Amazon Robotics Challenge 2016 and 2017.

Google Research Internships

I've done two internships at Google Research Zurich. In 2015 I worked on LSTMs for handwriting recognition. In 2016 I built a program that generated videos by learning how to stitch different video segments in an unsupervised way.