Avatar

Eleni Triantafillou

Ph.D. candidate
Supervised by Raquel Urtasun and Richard Zemel
University of Toronto
Machine Learning Group
Vector Institute

[eleni at cs dot toronto dot edu]

CV

Github

Scholar

Twitter

LinkedIn





About me

My research is centered around the pursuit of intelligent systems that are flexible and adaptable. I'm currently excited about:

  • Building systems that can effectively ingest large amounts of diverse data towards the goal of solving new learning tasks from unseen datasets or domains. We recently created the Meta-Dataset benchmark to facilitate research in this direction.
  • Understanding the nature of representations that enable few-shot learning. What is the role of meta-learning / episodic training towards that goal?
    • We studied a more challenging context-dependent variant of few-shot classification, and discovered that unsupervised learning is crucial for that problem.
    • Our ongoing work investigates the role of the episodic training paradigm through the lens of shot-specialization.

I also enjoy food blogging, dance (recently hip hop and contemporary), piano playing, singing and song writing.

I am graduating in the summer of 2021 and on the job market for research scientist or postdoc positions.

Publications

Learning a Universal Template for Few-shot Dataset Generalization.
Eleni Triantafillou, Hugo Larochelle, Richard Zemel, Vincent Dumoulin. ICML 2021.


Meta-dataset: A dataset of datasets for learning to learn from few examples.
Eleni Triantafillou, Tyler Zhu, Vincent Dumoulin, Pascal Lamblin, Kelvin Xu, Ross Goroshin, Carles Gelada, Kevin Swersky, Pierre-Antoine Manzagol, Hugo Larochelle. ICLR 2020. [paper].


Meta Learning for Semi-Supervised Few-Shot Classification.
Mengye Ren, Eleni Triantafillou*, Sachin Ravi*, Jake Snell, Kevin Swersky, Joshua B. Tenenbaum, Hugo Larochelle, Richard Zemel. ICLR 2018. [paper]


Few-Shot Learning Through an Information Retrieval Lens.
Eleni Triantafillou, Richard Zemel and Raquel Urtasun. NeurIPS 2017. [paper]


Non-Deterministic Planning with Temporally Extended Goals: LTL over finite and infinite traces.
Alberto Camacho, Eleni Triantafillou, Christian Muise, Jorge Baier, and Sheila McIlraith. AAAI, 2017. [paper]

Workshop papers and preprints

Flexible Few-Shot Learning with Contextual Similarity.
Mengye Ren*, Eleni Triantafillou*, Kuan-Chieh Wang*, James Lucas*, Jake Snell, Xaq Pitkow, Andreas S. Tolias, Richard Zemel. 2020. [paper].


Learning Flexible Classifiers with Shot-CONditional Episodic (SCONE) Training.
Eleni Triantafillou, Vincent Dumoulin, Hugo Larochelle, Richard Zemel. Meta-Learning workshop at NeurIPS 2020. [paper].


Few-shot Out-of-Distribution Detection.
Kuan-Chieh Wang, Paul Vicol, Eleni Triantafillou, Richard Zemel. UDL workshop at ICML 2020 (spotlight). [paper].


Few-shot Learning for Free by Modelling Global Class Structure.
Xuechen Li*, Will Grathwohl*, Eleni Triantafillou*, David Duvenaud, Richard Zemel. Meta-learning workshop at NeurIPS 2018. [paper].


Towards Generalizable Sentence Embeddings
Eleni Triantafillou, Jamie Ryan Kiros, Raquel Urtasun, Richard Zemel.
1st Workshop on Representation Learning for NLP at ACL 2016. [paper]


A Unifying Framework for Planning with LTL and Regular Expressions
Eleni Triantafillou, Jorge A. Baier, Sheila A. McIlraith.
MOCHAP workshop at ICAPS 2015. [paper]

Reviewing

  • NeurIPS: 2018 (top 10%), 2019, 2020 (top 10%).
  • ICML: 2019 (top 5%), 2020, 2021 (expert reviewer) .
  • ICLR: 2019, 2020, 2021 (outstanding reviewer)
  • CVPR: 2021
  • UAI: 2018
  • IROS: 2021
  • PC member for workshops: S2D-OLAD at ICLR 2021, Meta-Learning at NeurIPS: 2020 (senior reviewer), 2018 and 2017, AMTL at ICML 2019, LLD at ICLR 2019, LLD at NeurIPS 2017, WiML at NeurIPS 2017, WiCV at CVPR 2018 and 2021