Data Skeptic

Kyle met up with Damian Brady at MS Ignite 2019 to discuss machine learning operations.

The modern deep learning approaches to natural language processing are voracious in their demands for large corpora to train on.  Folk wisdom estimates used to be around 100k documents were required for effective training.  The availability of broadly trained, general-purpose models like BERT has made it possible to do transfer learning to achieve novel results on much smaller corpora. Thanks to these advancements, an NLP researcher might get value out of fewer examples since they can use the transfer learning to get a head start and focus on learning the nuances of the language specifically relevant to the task at hand.  Thus, small specialized corpora are both useful and practical to create. In this episode, Kyle speaks with Mor Geva, lead author on the recent paper Are We Modeling the Task or the Annotator? An Investigation of Annotator Bias in Natural Language Understanding Datasets, which explores some unintended consequences of the typical procedure followed for generating corpora. Source code for the paper available here: https://github.com/mega002/annotator_bias  

While at MS Build 2019, Kyle sat down with Lance Olson from the Applied AI team about how tools like cognitive services and cognitive search enable non-data scientists to access relatively advanced NLP tools out of box, and how more advanced data scientists can focus more time on the bigger picture problems.

Manuel Mager joins us to discuss natural language processing for low and under-resourced languages.  We discuss current work in this area and the Naki Project which aggregates research on NLP for native and indigenous languages of the American continent.

GPT-2 is yet another in a succession of models like ELMo and BERT which adopt a similar deep learning architecture and train an unsupervised model on a massive text corpus. As we have been covering recently, these approaches are showing tremendous promise, but how close are they to an AGI?  Our guest today, Vazgen Davidyants wondered exactly that, and have conversations with a Chatbot running GPT-2.  We discuss his experiences as well as some novel thoughts on artificial intelligence.

Rajiv Shah attempted to reproduce an earthquake-predicting deep learning model.  His results exposed some issues with the model.  Kyle and Rajiv discuss the original paper and Rajiv's analysis.

Allyson Ettinger joins us to discuss her work in computational linguistics, specifically in exploring some of the ways in which the popular natural language processing approach BERT has limitations.

Omer Levy joins us to discuss "SpanBERT: Improving Pre-training by Representing and Predicting Spans". https://arxiv.org/abs/1907.10529

Tim Niven joins us this week to discuss his work exploring the limits of what BERT can do on certain natural language tasks such as adversarial attacks, compositional learning, and systematic learning.

Kyle pontificates on how impressed he is with BERT.