From early school days, STEM students are often told that it is okay if they struggle with creative writing, or even English in general – after all you could never use mathematics to understand the intricacies of a language, right? In this talk, Nikita will explain how we can translate from the nuanced world of natural language to the beautifully simple world of vectors – with some help from our childhood friends ELMo and BERT – using natural language processing (NLP). She will provide some demonstrations of NLP in action, and later discuss some of the most common issues faced in the field.

No prerequisites are required.

Links mentioned in the video:
4:06 Linguistic Regularities in Continuous Space Word Representations - https://www.aclweb.org/anthology/N13-...
5:24 GLUE Benchmark - https://gluebenchmark.com/leaderboard
7:08 Evolution Strategies as a Scalable Alternative to Reinforcement Learning - https://openai.com/blog/evolution-str...
15:19 On the difficulty of training recurrent neural networks - http://proceedings.mlr.press/v28/pasc...
16:26 Attention Is All You Need - https://papers.nips.cc/paper/7181-att...
29:39 Explaining and Harnessing Adversarial Examples - https://arxiv.org/pdf/1412.6572.pdf
30:49 Robust Physical-World Attacks on Deep Learning Visual Classification - https://arxiv.org/pdf/1707.08945.pdf
33:45 Adversarial example using FGSM - https://www.tensorflow.org/tutorials/...

Credits to our friend John Conacher for the memes.

--------------------------------------------------------------------------------
Subscribe:    / @thearchimedeans7041  
Our website: https://archim.soc.srcf.net/
Our Facebook page:   / archimedeans  
Join our mailing list: https://lists.cam.ac.uk/mailman/listi...
Join our Microsoft Teams: http://forms.gle/tGHnN7XhEUVCeRgr6