Held Virtually – event open to Vector researchers and industry sponsors only

 

Register VIEW SEPT 15 AGENDA

 

September 16: AGENDA

Opening Remarks

10:00 am – 10:10 am

 

 

Keynote Presentation: AI at Work – How AI is transforming knowledge work

10:10 am – 10:40 am

In this talk, I will provide an overview of how AI in general and NLP in particular are transforming knowledge work and will share case studies from the Legal and News industries. I will also discuss our collaboration with the Vector Institute and other partner organizations on Deep Learning and the impact of deep learning on how Thomson Reuters develops AI-powered applications. I will conclude the talk with lessons learned from over 25 years of building machine learning and NLP applications and share my perspective on some of the opportunities that lie ahead.

 

 

Keynote Presentation: Say ‘ah’: Speech and language in medicine

10:40 am – 11:00 am

The healthcare industry is rapidly acquiring data, but organizing and using those data remains a challenge. Speech and language data, in particular, have multiple peculiar aspects that need to be overcome, in order to really derive the potential benefit, and to make healthcare more efficient, more affordable, and less prone to error. I will present a few use cases of NLP in healthcare, and also highlight a few of the current hurdles to operating this exciting technology in practice

 

 

Project/Research Presentation

11:00 am – 11:30 am

Project Presentation: An Experimental Evaluation of Large NLP Models in the Biomedical Domain
With the growing amount of text in health data, there have been rapid advances in large pre-trained models that can be applied to a wide variety of biomedical tasks with minimal task-specific modifications. Emphasizing the cost of these models, which renders technical replication challenging, in this project we present experiments replicating BioBERT and further pre-training and fine-tuning in the biomedical domain. We also investigate the effectiveness of domain-specific and domain-agnostic pre-trained models across downstream biomedical NLP tasks. Our finding confirms that pre-trained models can be impactful in some downstream NLP tasks (QA and NER) in the biomedical domain.

 

 

Project Presentation: An Investigation of Transformer based model in Legal Texts
In this presentation we overview the results of some of the experiments we conducted to best customize the state-of-the-art contextualized transformer-based language models to address the specific characteristics of the legal domain texts.

 

 

Student Presentation: Examining the rhetorical capacities of neural language models
Language models are key components for building neural NLP systems in discourse contexts. How should we choose between BERT, GPT-2, and other language models? We measure their abilities to understand rhetorical signals in texts, using a specially designed method called RST-probe.

 

 

Panel Discussion: Business Impacts: What is the main risk to NLP or from NLP in the next 5 years?”

Moderator:

 

Panelists:

 

 

Networking and Poster Session

12 noon – 12:30 pm

 

Poster #1: Application of NLP in Emergency Medical Services Amrit Sehdev Queen’s University	Poster #2: Modelling Sentence Pairs via Reinforcement Learning: An Actor-Critic Approach to Learn the Irrelevant Words Mahtab Ahmed University of Western Ontario	Poster #3: SentenceMIM: A Latent Variable Language Model Micha Livne University of Toronto
Poster #4: Training without training data: Improving the generalizability of automated medical abbreviation disambiguation Marta Skreta University of Toronto	Poster #5: Explainability for deep learning text classifiers Diana Lucaci University of Ottawa	Poster #6: Identifying Clinical Terms in Medical Text Using Ontology-Guided Machine Learning Aryan Arbabi University of Toronto
Poster #7: Sharing is Caring: Exploring machine learning methods to facilitate medical imaging exchange using metadata only Joanna Pineda University of Toronto	Poster #8: GOBO: Quantizing Attention-Based NLP Models for Low Latency and Energy Efficient Inference Ali Hadi Zadeh University of Toronto	Poster #9: Improved knowledge distillation by utilizing backward pass knowledge in neural networks Aref Jafari University of Waterloo
Poster #10: How Nouns Surface as Verbs: Inference and Generation in Word Class Conversion Lei Yu University of Toronto	Poster #11: Informal Natural Language Processing: The Case of Slang Zhewei Sun University of Toronto	Poster #12: Applications of the Chinese Remainder Theorem in Word Embedding Compression and Arithmetic Patricia Thaine University of Toronto
Poster #13: Predicting change in Major Depressive Disorder symptoms based on topic modelling features from psychiatric notes: An exploratory analysis Marta Maslej Centre for Addiction and Mental Health	Poster #14: Non-Pharmaceutical Intervention Discovery with Topic Modeling Jonathan Smith Layer 6	Poster #15: Hurtful words: quantifying biases in clinical contextual word embeddings Haoran Zhang University of Toronto
Poster #16: Domain Specific Fine-tuning of Denoising Sequence-to-Sequence Models for Natural Language Summarization Matt Kalebic PwC/Deloitte	Poster #17: An Experimental Evaluation of Large NLP Models in the Biomedical Domain Faiza Khattak Khan Data Scientist, Manulife	Poster #18: An Investigation of Transformer based model in Legal Texts Shohreh Shaghaghian Research Scientist at Center for AI and Cognitive Computing, Thomson Reuters
Poster #19: Examining the rhetorical capacities of neural language models Zining Zhu PhD Student, University of Toronto	Poster #20: Using natural language processing to predict splenomegaly from >100,000 structured CT reports Karen Batch Queen’s University

 

Networking and Break

12:30 pm – 12:45 pm

 

Concurrent Workshops

12:45 pm – 1:45 pm

 

WS1: How to use Fairseq (Facebook AI Research Sequence-to-Sequence Toolkit) for your own NLP application
Fairseq library is a sequence modeling toolkit that allows researchers and developers to train custom models for translation, summarization, language modeling and other text generation tasks. In this short workshop we will walk participants through the basics of the Fairseq library. We will dive into their codebase and learn how to modify existing modules to create and keep track of new applications. The purpose of this workshop is to provide learning through demonstration and hands-on experience.

Level of workshop: Intermediate/Advanced

 

WS2: Question Answering Systems in Responding to COVID-19 Open Research Dataset Challenge
In response to the COVID-19 pandemic, the White House and a coalition of leading research groups have prepared the COVID-19 Open Research Dataset (CORD-19). In this workshop we demonstrate three Question Answering systems that were submitted to a Kaggle COVID-19 Open Research Dataset Challenge that could help the medical community develop answers to high priority scientific questions. The competition has launched to provide a chance for the machine learning research communities to employ advanced NLP methods to form a QA system for finding scientific answers for questions related to COVID-19. To do so, a dataset that is a collection of scholarly studies on the coronavirus group (i.e., referred to as the CORD-19) has been provided as a result of a collaboration between different research institutes such as Microsoft Research, the Allen Institute for AI, the National Library of Medicine at the NIH.

Level of workshop: Intermediate

 

Event to conclude

1:45 pm

 

Register