Highlights
- Started new chapter at Google!
- Sparse Meta Networks, a follow-up work on Meta Networks is on arxiv. Sparse-MetaNet improves non-adaptive Transformer-XL baseline results on WikiText-103 and Enwiki8.
- Metalearned Neural Memory got accepted at NeurIPS 2019!
- We successfully organized DLUB 2019 - a summer school on deep learning. The theme this year was NLP.
- Selected for ICML travel award: I am very grateful to the International Machine Learning Society.
- Our paper, Meta Networks (MetaNet) is accepted at ICML 2017!
- My dissertation is available: Domain Knowledge-rich Biomedical Information Extraction Methods based on Deep Neural Networks.
- I got the chance to attend Deep Learning Summer School 2015 in Montreal. There I presented our poster: DeepText: End-to-end biomedical event extraction via deep learning and recursive projection model
Open source software
- DeepText: An end-to-end biomedical event extraction system based on deep learning and recursive projection model, written in Scala and Lua (using Torch).
- BANNER-CHEMDNER: A multi-domain named entity recognition system that can be used to identify chemical and drug mention, biomedical mention or disease mention from text, written in Java.
Selected Publications
Conference and workshop papers
- Rajarshi Das, Tsendsuren Munkhdalai, Xingdi Yuan, Adam Trischler, and Andrew McCallum. "Building Dynamic Knowledge Graphs from Text using Machine Reading Comprehension." ICLR, 2019
- Tsendsuren Munkhdalai, Xingdi Yuan, Soroush Mehri, and Adam Trischler. "Rapid Adaptation with Conditionally Shifted Neurons." ICML, 2018
- Tu Vu, Baotian Hu, Tsendsuren Munkhdalai, and Hong Yu. "Sentence Simplification with Memory-Augmented Neural Networks." NAACL, 2018
- John Lalor, Hao Wu, Tsendsuren Munkhdalai, and Hong Yu. "Understanding Deep Learning Performance through an Examination of Test Set Difficulty: A Psychometric Case Study." EMNLP, 2018
- Tsendsuren Munkhdalai and Hong Yu. "Meta networks." ICML, 2017 [code]
- Tsendsuren Munkhdalai and Hong Yu. "Reasoning with memory augmented neural networks for language comprehension." ICLR, 2017
- Tsendsuren Munkhdalai and Hong Yu. "Neural tree indexers for text understanding." EACL, 2017 [code]
- Tsendsuren Munkhdalai and Hong Yu. "Neural semantic encoders." EACL, 2017 [code]
- Tsendsuren Munkhdalai, John Lalor, and Hong Yu. "Citation analysis with neural attention models." Health Text Mining and Information Analysis, EMNLP workshop, 2016
- Tsendsuren Munkhdalai, Meijing Li, Khuyagbaatar Batsuren and Keun Ho Ryu. "Towards a unified named entity recognition system: disease mention identification." In Proceedings of the 6th International Conference on Bioinformatics Models, Methods and Algorithms, 2015 [code]
-
Tsendsuren Munkhdalai, Meijing Li, Khuyagbaatar Batsuren and Keun Ho Ryu. "BANNER-CHEMDNER: Incorporating domain knowledge in chemical and drug named entity recognition." In Proceedings of the 4th BioCreative Challenge Evaluation Workshop vol. 2, 2013 [code]
Placed 6th (out of 23) in the CHEMDNER chemical document indexing subtask and 8th (out of 26) in the Chemical entity mention recognition subtask - Tsendsuren Munkhdalai, Meijing Li, Erdenetuya Namsrai, Oyun-Erdene Namsrai and Keun Ho Ryu. "BFSM: Finite state machine learned as named boundary definer for bio named entity recognition." In Proceedings of the 3rd IEEE International Conference on Awareness Science and Technology, 2011.
Refereed journals
- Lkhagvadorj Munkhdalai, Tsendsuren Munkhdalai, Oyun-Erdene Namsrai, Jong Yun Lee, and Keun Ho Ryu. "An Empirical Comparison of Machine-Learning Methods on Bank Client Credit Assessments." Sustainability, 2019
- Tsendsuren Munkhdalai, Feifan Liu, and Hong Yu. "Clinical relation extraction toward drug safety surveillance using electronic health record narratives: classical learning versus deep learning." JMIR public health and surveillance, 2018
- Tsendsuren Munkhdalai, Meijing Li, Khuyagbaatar Batsuren, Hyeon Ah Park, Nak Hyeon Choi and Keun Ho Ryu. "Incorporating domain knowledge in chemical and biomedical named entity recognition with word representations." Journal of Cheminformatics, 2015 [link|code]
- Tsendsuren Munkhdlai, Oyun-Erdene Namsrai and Keun Ho Ryu. "Self-training significance space of support vectors for imbalanced biomedical event data." BMC Bioinformatics, 2015 (accepted for a special issue of BIOT 2014) [code]
- Meijing Li, Tsendsuren Munkhdalai, Xiuming Yu and Keun Ho Ryu. "A Novel Approach for Protein-Named Entity Recognition and Protein-Protein Interaction Extraction." Mathematical Problems in Engineering, 2015
- Martin Krallinger, Obdulia Rabal, Florian Leitner, Miguel Vazquez, David Salgado, Zhiyong Lu, Robert Leaman, Yanan Lu, Donghong Ji, Daniel M Lowe, Roger A Sayle, Riza Batista-Navarro, Rafal Rak, Torsten Huber, Tim Rocktäschel, Sérgio Matos, David Campos, Buzhou Tang, Hua Xu, Tsendsuren Munkhdalai, Keun Ryu, SV Ramanan, Senthil Nathan, Slavko Žitnik, Marko Bajec, Lutz Weber, Matthias Irmer, Saber A Akhondi, Jan A Kors, Shuo Xu, Xin An, Utpal Sikdar, Asif Ekbal, Masaharu Yoshioka, Thaer M Dieb, Miji Choi, Karin Verspoor, Madian Khabsa, C Giles, Hongfang Liu, Komandur Ravikumar, Andre Lamurias, Francisco M Couto, Hong-Jie Dai, Richard Tsai, Caglar Ata, Tolga Can, Anabel Usié, Rui Alves, Isabel Segura-Bedmar, Paloma Martínez, Julen Oyarzabal and Alfonso Valencia. "The CHEMDNER corpus of chemicals and drugs and its annotation principles." Journal of Cheminformatics, 2015 [link]
- Erdenetuya Namsrai, Tsendsuren Munkhdalai, Meijing Li, Jung-Hoon Shin, Oyun- Erdene Namsrai and Keun Ho Ryu. "A Feature Selection-based Ensemble Method for Arrhythmia Classification." Journal of Information Processing Systems, 2013
- Tsendsuren Munkhdalai, Meijing Li, Unil Yun, Oyun-Erdene Namsrai and Keun Ho Ryu. "An active co-training algorithm for biomedical named-entity recognition." Journal of Information Processing Systems, 2012
Pre-prints and technical reports
- Tsendsuren Mukhdalai and Adam Trischler. "Metalearning with hebbian fast weights." arXiv preprint arXiv:1807.05076 (2018).
- Tsendsuren Munkhdalai and Keun Ho Ryu. "DeepText: End-to-end biomedical event extraction via deep learning and recursive projection model." Technical report and now part of my thesis [code]