Vitaly Kuznetsov

I am an Algo Developer at Hudson River Trading. I lead a team of researchers developing models and algorithms for trading on financial markets.

Prior to joining HRT, I was a Senior Research Scientist at Google AI. While at Google, I have developed machine learning solutions for a variety of problems including time series forecasting for Ads traffic, AutoML and machine translation.

I completed a Ph.D. degree in mathematics at the Courant Institute of Mathematical Sciences, where I worked with Professor Mehryar Mohri. Before coming to the Courant Institute, I received Bachelor's and Master's degrees in mathematics from the University of Toronto.

Selected Academic Activity

• Open source software: AdaNet. Also, see paper and Google AI blog.
• Kaggle competition for time series forecasting.
• A series of workshops on anomaly detection and time series analysis: ICML 2019, NIPS 2017, ICML 2017, NIPS 2016, ICML 2016, NIPS 2015.
• Tutorial on Theory and algorithms for non-stationary time series forecasting at NIPS 2016 [Video].
• Area chair for major ML conference: Neurips (since 2016), ICML (since 2021), ICLR (since 2021).

Publications (Google Scholar)

1. Vitaly Kuznetsov and Mehryar Mohri.
   Discrepancy-based theory and algorithms for forecasting non-stationary time series.
   Annals of Mathematics and Artificial Intelligence. 86(1):1-33, 2020.
2. Vitaly Kuznetsov and Zelda Mariet.
   Foundations of Sequence-to-Sequence Modeling for Time Series.
   In Twenty-Second Conference on Artificial Intelligence and Statistics (AISTATS 2019). Naha, Okinawa, Japan, April 2019.
3. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri, Holakou Rahmanian and Manfred Warmuth.
   Online non-additive path learning under full and partial information.
   In Proceedings of the 30th International Conference on Algorithmic Learning Theory (ALT 2019). Chicago, USA, March 2019.
4. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri, Dmitry Storcheus, and Scott Yang.
   Efficient gradient computation for structured output learning with rational ad tropical losses.
   In Advances in Neural Information Processing Systems (NeurIPS 2018). Montreal, Canada, December 2018.
5. Vitaly Kuznetsov and Mehryar Mohri.
   Discriminative state-space models.
   In Advances in Neural Information Processing Systems (NIPS 2017). Long Beach, CA, December 2017.
6. Corinna Cortes, Xavier Gonzalvo, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   AdaNet: Adaptive structural learning of artificial neural networks.
   In Proceedings of the 34st International Conference on Machine Learning (ICML 2017). Sydney, Australia, August 2017.
7. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   Structured prediction theory based on factor graph complexity.
   In Advances in Neural Information Processing Systems (NIPS 2016). Barcelona, Spain, December 2016.
8. Vitaly Kuznetsov, Hank Liao, Mehryar Mohri, Michael Riley and Brian Roark.
   Learning N-gram language models from uncertain data.
   In Proceedings of the 17th Annual Conference of the International Speech Communication Association (Interspeech 2016). San Francisco, USA, September 2016.
9. Vitaly Kuznetsov and Mehryar Mohri.
   Time series prediction and online learning.
   In Proceedings of The 29th Annual Conference on Learning Theory (COLT 2016). New York, USA, June 2016.
10. Vitaly Kuznetsov and Mehryar Mohri.
    Generalization bounds for non-stationary mixing processes.
    Machine Learning Journal, 106:1-25, 2016.
11. Corinna Cortes, Prasoon Goyal, Vitaly Kuznetsov and Mehryar Mohri.
    Kernel extraction via voted risk minimization.
    Journal of Machine Learning Research (JMLR), 44:72-89, 2015.
12. Vitaly Kuznetsov and Mehryar Mohri.
    Learning theory and algorithms for forecasting non-stationary time series.
    In Advances in Neural Information Processing Systems (NIPS 2015). Montreal, Canada, December 2015.
    (full oral, top 15 papers out of 1838 submissions)
13. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri and Manfred Warmuth.
    On-line learning algorithms for path experts with non-additive losses.
    In Proceedings of The 28th Annual Conference on Learning Theory (COLT 2015). Paris, France, July 2015.
14. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri and Umar Syed.
    Structural maximum entropy models.
    In Proceedings of the 32st International Conference on Machine Learning (ICML 2015). Lille, France, July 2015.
15. Vitaly Kuznetsov, Mehryar Mohri and Umar Syed.
    Multi-class deep boosting.
    In Advances in Neural Information Processing Systems (NIPS 2014). Montreal, Canada, December 2014.
16. Vitaly Kuznetsov and Mehryar Mohri.
    Generalization bounds for time series prediction with non-stationary processes.
    In Proceedings of the 25th International Conference on Algorithmic Learning Theory (ALT 2014). Bled, Slovenia, October 2014.
17. Corinna Cortes, Vitaly Kuznetsov and Mehryar Mohri.
    Learning ensembles of structured prediction rules.
    In Proceedings of the 52nd Annual Meeting of Association for Computational Linguistics (ACL 2014). Baltimore, USA, June 2014.
18. Corinna Cortes, Vitaly Kuznetsov and Mehryar Mohri.
    Ensemble methods for structured prediction.
    In Proceedings of the 31st International Conference on Machine Learning (ICML 2014). Beijing, China, June 2014.
19. Abraham Isgur, Vitaly Kuznetsov, Mustazee Rahman and Stephen Tanny.
    Nested recursions, simultaneous parameters and tree superpositions.
    Electronic Journal of Combinatorics. Volume 21, Issue 1, 2014.
20. Abraham Isgur, Vitaly Kuznetsov and Stephen Tanny.
    A combinatorial approach for solving certain nested recursions with non-slow solutions.
    Journal of Difference Equations and Applications, Volume 19, Issue 4, 2013.
21. Rafal Drabek, Abraham Isgur, Vitaly Kuznetsov and Stephen Tanny.
    Sums of ceiling functions solve nested recursions
    Journal of Difference Equations and Applications, Volume 18, Issue 12, 2012.
22. Abraham Isgur, Vitaly Kuznetsov and Stephen Tanny.
    Nested recursions with ceiling function solutions
    Journal of Difference Equations and Applications, Volume 18, Issue 6, 2012.

• Oren Anava, Azadeh Khaleghi, Marco Cuturi, Vitaly Kuznetsov, Alexander Rakhlin (Editors).
  Proceedings of Machine Learning Research: NIPS 2016 Time Series Workshop. Volume 55.

1. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri, Dmitry Storcheus, and Scott Yang.
   Efficient gradient computation for structured output learning with rational ad tropical losses.
   In NeurIPS 2018 Workshop on Visually Grounded Interaction and Language. Montreal, Canada, December 2018.
2. Corinna Cortes, Xavier Gonzalvo, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   AdaNet: Adaptive Structural Learning of Artificial Neural Networks.
   In ICML 2017 Workshop on Principled approaches to deep learning. Sydney, Australia, August 2017.
3. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   Measuring Learnability in Structured Prediction using Factor Graph Complexity.
   In NIPS 2016 Workshop on Learning in High Dimensions with Structure. Barcelona, Spain, December 2016.
4. Corinna Cortes, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   A Theoretical Framework for Structured Prediction using Factor Graph Complexity.
   In NIPS 2016 Workshop on Extreme Classification. Barcelona, Spain, December 2016.
5. Corinna Cortes, Xavier Gonzalvo, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   AdaNet: Adaptive Structural Learning of Artificial Neural Networks.
   In NIPS 2016 Workshop on Adaptive and Scalable Nonparametric Methods in Machine Learning. Barcelona, Spain, December 2016.
6. Corinna Cortes, Xavier Gonzalvo, Vitaly Kuznetsov, Mehryar Mohri and Scott Yang.
   AdaNet: Adaptive Structural Learning of Artificial Neural Networks.
   In NIPS 2016 Workshop on Efficient Methods for Deep Neural Networks. Barcelona, Spain, December 2016.
7. Vitaly Kuznetsov, Mehryar Mohri and Umar Syed.
   Rademacher complexity margin bounds for learning with a large number of classes.
   In ICML 2015 Workshop on Extreme Classification. Lille, France, July 2015.
8. Corinna Cortes, Vitaly Kuznetsov and Mehryar Mohri.
   Boosting ensembles of structured prediction rules.
   In NIPS 2014 Workshop on Modern Machine Learning and Natural Language Processing. Montreal, Canada, December 2014.
9. Corinna Cortes, Vitaly Kuznetsov and Mehryar Mohri.
   On-line learning approach to ensemble methods for structured prediction.
   In NIPS 2014 Workshop on Representation and Learning Methods Complex Outputs. Montreal, Canada, December 2014.
10. Vitaly Kuznetsov and Mehryar Mohri.
    Forecasting non-stationary time series: from theory to algorithms.
    In NIPS 2014 Workshop on Transfer and Multi-task Learning. Montreal, Canada, December 2014.

• AdaNet: Adaptive Structural Learning of Artificial Neural Networks. ICML 2017, Sydney, Australia, August 2017. Video (password: icml2017)
• NIPS 2016 Tutorial: Learning Theory and Algorithms for Time Series Prediction. NIPS 2016, Barcelona, Spain, December 2016. Video Slides
• Time Series Prediction and Online Learning. COLT 2016, New York, USA, June 2016. Video
• Learning Theory and Algorithms for Time Series Prediction. NIPS 2015, Montreal, Canada, December 2015. Video
• Ensemble methods for structured prediction.
  ICML Workshop on Features and Structures 2015 (invited lecture), Lille, France, July 2015. Slides
• Structural maximum entropy models. ICML 2015, Lille, France, July 2015. Video
• On-line learning algorithms for path experts with non-additive losses. COLT 2015, Paris, France, July 2015. Video
• On-line learning approach to ensemble methods for structured prediction.
  NIPS 2014 Workshop on Representation and Learning Methods Complex Outputs, Montreal, Canada, December 2014. Video
• Ensemble methods for structured prediction. ICML 2014, Beijing, China, June 2014. Video