Iulian Serban

Research

My research is about machines that converse and teach. During my PhD at Mila and the Université de Montréal (2014–2019), supervised by Yoshua Bengio and Joelle Pineau, I worked on generative neural network models for dialogue, on how to evaluate conversational systems, and on reinforcement learning for chatbots. Before that I did an MSc in machine learning at University College London, supervised by Peter Dayan and David Silver.

At Korbit, we published a dozen papers on AI for education: Socratic tutoring, generated learning content and adaptive learning paths. At LawZero I worked on AI safety research, and we published three papers.

I won an Outstanding Long Paper Award at ACL 2017 for Towards an Automatic Turing Test: Learning to Evaluate Dialogue Responses, and 2nd Prize in the NeurIPS Demonstration Track for A Deep Reinforcement Learning Chatbot. In all, I’ve published 40+ papers, co-authored two books, received five awards and hold a US patent on systems and methods for reviewing code.

Selected papers

  1. Building end-to-end dialogue systems using generative hierarchical neural network modelsIulian Serban, Alessandro Sordoni, Yoshua Bengio, Aaron Courville and Joelle PineauAAAI, 2016arXiv
  2. A hierarchical latent variable encoder-decoder model for generating dialoguesIulian Serban, Alessandro Sordoni, Ryan Lowe, Laurent Charlin, Joelle Pineau, Aaron Courville and Yoshua BengioAAAI, 2017arXiv
  3. The Ubuntu dialogue corpus: A large dataset for research in unstructured multi-turn dialogue systemsRyan Lowe, Nissan Pow, Iulian Vlad Serban and Joelle PineauSIGDIAL, 2015
  4. Towards an automatic Turing test: Learning to evaluate dialogue responsesRyan Lowe, Michael Noseworthy, Iulian Vlad Serban, Nicolas Angelard-Gontier, Yoshua Bengio and Joelle PineauACL, 2017Outstanding Long Paper AwardPaper arXiv
  5. A deep reinforcement learning chatbotIulian V. Serban, Chinnadhurai Sankar, Mathieu Germain, Saizheng Zhang, Zhouhan Lin, Sandeep Subramanian, Taesup Kim, Michael Pieper, Sarath Chandar, Nan Rosemary Ke and othersarXiv, 20172nd Prize in the NeurIPS Demonstration TrackarXiv
  6. A large-scale, open-domain, mixed-interface dialogue-based ITS for STEMIulian Vlad Serban, Varun Gupta, Ekaterina Kochmar, Dung D. Vu, Robert Belfer, Joelle Pineau, Aaron Courville, Laurent Charlin and Yoshua BengioAIED, 2020
  7. The Scientist AI: Safe by Design, by Not DesiringDamiano Fornasiere, Oliver Richardson, Gaël Gendron, Iulian Serban and Yoshua BengioLawZero technical report, 2026

All papers

47 papers, newest first. Citation counts are on Google Scholar.

2026

  1. The Scientist AI: Safe by Design, by Not DesiringDamiano Fornasiere, Oliver Richardson, Gaël Gendron, Iulian Serban and Yoshua BengioLawZero technical report
  2. Safety from Honesty in a Disinterested AI PredictorYoshua Bengio, Oliver Richardson, Tomáš Gavenčiak, Michael Cohen, Rory Svarc, Damiano Fornasiere, Gael Gendron, David Hyland, Aton Kamanda, Adam Oberman and othersarXivarXiv
  3. PyINE: A Framework for Scalable Elicitation and Oversight via Code ExecutionPierre-Luc St-Charles, Alessandro Palmas, Damiano Fornasiere, Storm Lei, Mirko Bronzi, Jean-Pierre Falet, Iulian Serban and Yoshua BengioarXivarXiv

2024

  1. How teachers can use large language models and Bloom’s taxonomy to create educational quizzesSabina Elkins, Ekaterina Kochmar, Jackie C. K. Cheung and Iulian SerbanAAAI

2023

  1. How useful are educational questions generated by large language models?Sabina Elkins, Ekaterina Kochmar, Iulian Serban and Jackie C. K. CheungAIED

2022

  1. Automated data-driven generation of personalized pedagogical interventions in intelligent tutoring systemsEkaterina Kochmar, Dung Do Vu, Robert Belfer, Varun Gupta, Iulian Vlad Serban and Joelle PineauInternational Journal of Artificial Intelligence in Education
  2. A new era: Intelligent tutoring systems will transform online learning for millionsFrancois St-Hilaire, Dung Do Vu, Antoine Frau, Nathan Burns, Farid Faraji, Joseph Potochny, Stephane Robert, Arnaud Roussel, Selene Zheng, Taylor Glazier and othersarXivarXiv
  3. Few-shot question generation for personalized feedback in intelligent tutoring systemsDevang Kulshreshtha, Muhammad Shayan, Robert Belfer, Siva Reddy, Iulian Vlad Serban and Ekaterina KochmarPAIS 2022, co-located with IJCAI-ECAI
  4. Question Personalization in an Intelligent Tutoring SystemSabina Elkins, Ekaterina Kochmar, Robert Belfer, Iulian Serban and Jackie C. K. CheungAIED
  5. Raising student completion rates with adaptive curriculum and contextual banditsRobert Belfer, Ekaterina Kochmar and Iulian Vlad SerbanAIED

2021

  1. Deep discourse analysis for generating personalized feedback in intelligent tutor systemsMatt Grenander, Robert Belfer, Ekaterina Kochmar, Iulian V. Serban, François St-Hilaire and Jackie C. K. CheungAAAI
  2. A comparative study of learning outcomes for online learning platformsFrancois St-Hilaire, Nathan Burns, Robert Belfer, Muhammad Shayan, Ariella Smofsky, Dung Do Vu, Antoine Frau, Joseph Potochny, Farid Faraji, Vincent Pavero and othersAIED
  3. Back-training excels self-training at unsupervised domain adaptation of question generation and passage retrievalDevang Kulshreshtha, Robert Belfer, Iulian Vlad Serban and Siva ReddyEMNLP

2020

  1. The bottleneck simulator: A model-based deep reinforcement learning approachIulian Vlad Serban, Chinnadhurai Sankar, Michael Pieper, Joelle Pineau and Yoshua BengioJournal of Artificial Intelligence Research
  2. Automated personalized feedback improves learning gains in an intelligent tutoring systemEkaterina Kochmar, Dung Do Vu, Robert Belfer, Varun Gupta, Iulian Vlad Serban and Joelle PineauAIED
  3. A large-scale, open-domain, mixed-interface dialogue-based ITS for STEMIulian Vlad Serban, Varun Gupta, Ekaterina Kochmar, Dung D. Vu, Robert Belfer, Joelle Pineau, Aaron Courville, Laurent Charlin and Yoshua BengioAIED

2019

  1. The second conversational intelligence challenge (ConvAI2)Emily Dinan, Varvara Logacheva, Valentin Malykh, Alexander Miller, Kurt Shuster, Jack Urbanek, Douwe Kiela, Arthur Szlam, Iulian Serban, Ryan Lowe and othersThe NeurIPS'18 Competition: From Machine Learning to Intelligent Conversations (Springer)
  2. Representation learning for dialogue systemsIulian Vlad SerbanPhD thesis, Université de Montréal

2018

  1. A survey of available corpora for building data-driven dialogue systems: The journal versionIulian Vlad Serban, Ryan Lowe, Peter Henderson, Laurent Charlin and Joelle PineauDialogue & Discourse
  2. A deep reinforcement learning chatbot (short version)Iulian V. Serban, Chinnadhurai Sankar, Mathieu Germain, Saizheng Zhang, Zhouhan Lin, Sandeep Subramanian, Taesup Kim, Michael Pieper, Sarath Chandar, Nan Rosemary Ke and othersarXivarXiv
  3. A dataset of topic-oriented human-to-chatbot dialoguesVarvara Logacheva, Mikhail Burtsev, Valentin Malykh, Vadim Poluliakh, Alexander Rudnicky, Iulian Serban, Ryan Lowe, Shrimai Prabhumoye, Alan W. Black and Yoshua Bengio
  4. Introduction to NIPS 2017 Competition TrackSergio Escalera, Markus Weimer, Mikhail Burtsev, Valentin Malykh, Varvara Logacheva, Ryan Lowe, Iulian Vlad Serban, Yoshua Bengio, Alexander Rudnicky, Alan W. Black and othersThe NIPS'17 Competition: Building Intelligent Systems (Springer)
  5. The first conversational intelligence challengeMikhail Burtsev, Varvara Logacheva, Valentin Malykh, Iulian Vlad Serban, Ryan Lowe, Shrimai Prabhumoye, Alan W. Black, Alexander Rudnicky and Yoshua BengioThe NIPS'17 Competition: Building Intelligent Systems (Springer)
  6. The RLLChatbot: A solution to the ConvAI challengeNicolas Gontier, Koustuv Sinha, Peter Henderson, Iulian Serban, Michael Noseworthy, Prasanna Parthasarathi and Joelle PineauarXivarXiv

2017

  1. A hierarchical latent variable encoder-decoder model for generating dialoguesIulian Serban, Alessandro Sordoni, Ryan Lowe, Laurent Charlin, Joelle Pineau, Aaron Courville and Yoshua BengioAAAIarXiv
  2. Multiresolution recurrent neural networks: An application to dialogue response generationIulian Serban, Tim Klinger, Gerald Tesauro, Kartik Talamadupula, Bowen Zhou, Yoshua Bengio and Aaron CourvilleAAAI
  3. Training end-to-end dialogue systems with the Ubuntu dialogue corpusRyan Lowe, Nissan Pow, Iulian Vlad Serban, Laurent Charlin, Chia-Wei Liu and Joelle PineauDialogue & Discourse
  4. Towards an automatic Turing test: Learning to evaluate dialogue responsesRyan Lowe, Michael Noseworthy, Iulian Vlad Serban, Nicolas Angelard-Gontier, Yoshua Bengio and Joelle PineauACLOutstanding Long Paper AwardPaper arXiv
  5. Piecewise latent variables for neural variational text processingIulian Vlad Serban, Alexander G. Ororbia, Joelle Pineau and Aaron CourvilleEMNLP
  6. A deep reinforcement learning chatbotIulian V. Serban, Chinnadhurai Sankar, Mathieu Germain, Saizheng Zhang, Zhouhan Lin, Sandeep Subramanian, Taesup Kim, Michael Pieper, Sarath Chandar, Nan Rosemary Ke and othersarXiv2nd Prize in the NeurIPS Demonstration TrackarXiv
  7. The octopus approach to the Alexa competition: A deep ensemble-based socialbotIulian V. Serban, Chinnadhurai Sankar, Saizheng Zhang, Zhouhan Lin, Sandeep Subramanian, Taesup Kim, Sarath Chandar, Nan Rosemary Ke, Sai Rajeswar, Alexandre de Brebisson and othersAlexa Prize Proceedings

2016

  1. Building end-to-end dialogue systems using generative hierarchical neural network modelsIulian Serban, Alessandro Sordoni, Yoshua Bengio, Aaron Courville and Joelle PineauAAAIarXiv
  2. How not to evaluate your dialogue system: An empirical study of unsupervised evaluation metrics for dialogue response generationChia-Wei Liu, Ryan Lowe, Iulian Vlad Serban, Mike Noseworthy, Laurent Charlin and Joelle PineauEMNLP
  3. Generating factoid questions with recurrent neural networks: The 30M factoid question-answer corpusIulian Vlad Serban, Alberto García-Durán, Çağlar Gülçehre, Sungjin Ahn, Sarath Chandar, Aaron Courville and Yoshua BengioACL
  4. Generative deep neural networks for dialogue: A short reviewIulian Vlad Serban, Ryan Lowe, Laurent Charlin and Joelle PineauarXivarXiv
  5. On the evaluation of dialogue systems with next utterance classificationRyan Lowe, Iulian Vlad Serban, Michael Noseworthy, Laurent Charlin and Joelle PineauSIGDIAL
  6. Theano: A Python framework for fast computation of mathematical expressionsThe Theano Development Team, Rami Al-Rfou, Guillaume Alain, Amjad Almahairi, Christof Angermueller, Dzmitry Bahdanau, Nicolas Ballas, Frédéric Bastien, Justin Bayer, Anatoly Belikov and othersarXivarXiv

2015

  1. The Ubuntu dialogue corpus: A large dataset for research in unstructured multi-turn dialogue systemsRyan Lowe, Nissan Pow, Iulian Vlad Serban and Joelle PineauSIGDIAL
  2. Incorporating unstructured textual knowledge sources into neural dialogue systemsRyan Lowe, Nissan Pow, Iulian Serban, Laurent Charlin and Joelle PineauNIPS Workshop on Machine Learning for Spoken Language Understanding
  3. Text-based speaker identification for multi-participant open-domain dialogue systemsIulian V. Serban and Joelle PineauNIPS Workshop on Machine Learning for Spoken Language Understanding

2014

  1. Selected Graphical Models and Their Applications in Data AnalysisIulian Vlad Serban
  2. Prediction of changes in the stock market using Twitter and sentiment analysisIulian Vlad Serban, David Sierra González and Xuyang WuJournal of engineering and technology
  3. Learning to Play Chess from ScratchIulian Vlad Serban, David Barber and Peter Dayan
  4. Implicit Affective TaggingIulian Vlad Serban

2012

  1. Maximum likelihood learning and inference in conditional random fieldsIulian Vlad SerbanBachelor's thesis, University of Copenhagen
  2. Seizure Detection Challenge The Fitzgerald team solutionVincent Adam, Joana Soldado-Magraner, Wittawat Jitkritum, Heiko Strathmann, Balaji Lakshminarayanan, Alessandro Davide Ialongo, Gergő Bohner, Ben Dongsung Huh, Lea Goetz, Shaun Dowling and othersMechanics in Medicine and Biology

2011

  1. Fattigdom i Danmark: En socioøkonomisk fattigdomsgrænseIulian Vlad SerbanLIS Working Paper Series

Patents

  1. Systems and methods for reviewing codeIulian Serban, Ariella Smofsky, Robert Belfer, Mathieu Germain, Antoine Frau, Kavita Ganeshan, Ekaterina Kochmar, Soner Yildirim and Arjun SudhakarUS Patent 12,340,191, 2025
  2. Systems and methods for reviewing codeIulian Serban, Ariella Smofsky, Robert Belfer, Mathieu Germain, Antoine Frau, Kavita Ganeshan, Ekaterina Kochmar, Soner Yildirim and Arjun SudhakarUS Patent App. 19/212,614, 2026