I am a research scientist at Google Berlin working on machine translation and other structured prediction problems for NLP. Simultaneously, I hold an honorary professor position at the Heidelberg University.
Before Google I was at Amazon and earlier at the Statistical NLP Group, lead by Prof. Stefan Riezler (where I'm still teaching ML courses), and even earlier at the LIMSI and the Orange Labs, R&D in France. I obtained a PhD in Computer Science and Artificial Intelligence from the IRTCITS research center in Kyiv for a thesis on randomized algorithms for locality-sensitive embeddings of the Levenstein edit distance.
- Paper accepted at LOD on sparse zero-order optimization.
- Paper accepted at IWSLT on learning to segment for NMT.
- I was appointed an honorary professor and teaching an imitation learning course in winter semester 2018/2019.
- I co-organized a shared task on learning machine translation systems from weak feedback at WMT'17 that ran from January till July 2017. Official results.
- J. Kreutzer, A. Sokolov. Learning to Segment Inputs for NMT Favors Character-Level Processing, Int. Workshop on Spoken Language Translation (IWSLT), 2018 [extended report]
- F. Hieber, T. Domhan, M. Denkowski, D. Vilar, A. Sokolov, A. Clifton, M. Post. Sockeye: A Toolkit for Neural Machine Translation, American Machine Translation Association (AMTA), 2018 [complete ArXiv version] [code]
- C. Lawrence, A. Sokolov, S. Riezler. Counterfactual Learning from Bandit Feedback under Deterministic Logging: A Case Study in Statistical Machine Translation, Empirical Methods in Natural Language Processing (EMNLP), Copenhagen, Denmark, 2017 [poster]
- A. Sokolov, J. Kreutzer, K. Sunderland, P. Danchenko, W. Szymaniak, H. Fürstenau, S. Riezler. A Shared Task on Bandit Learning for Machine Translation, Conference on Machine Translation (WMT), Copenhagen, Denmark, 2017 [slides]
- J. Kreutzer, A. Sokolov, S. Riezler. Bandit Structured Prediction for Neural Sequence-to-Sequence Learning, Association of Computational Linguistics (ACL), Vancouver, Canada, 2017 [poster]
- A. Sokolov, J. Kreutzer, C. Lo, S. Riezler. Stochastic Structured Prediction under Bandit Feedback, Neural Information Processing Systems (NIPS), Barcelona, Spain, 2016 [poster]
- A. Sokolov, J. Kreutzer, C. Lo, S. Riezler. Learning Structured Predictors from Bandit Feedback for Interactive NLP, Association of Computational Linguistics (ACL), Berlin, Germany, 2016 [slides] [code 1] [code 2]
- V. Boteva, D. Gholipour, A. Sokolov, S. Riezler. A Full-Text Learning to Rank Dataset for Medical Information Retrieval, European Conference on Information Retrieval (ECIR), Padova, Italy, 2016 [poster] [data] [bib]
Institut für Computerlinguistik,
Im Neuenheimer Feld 325, Room 107