nlp highlights

Temple

nlp highlights playlist

a lecture (with music) about scientific materialism and its pitfalls 

Manly P Hall Science Beyond Materialism  (mix)

Hi Crystal. This upload will help you prepare for the task at hand.  There are 2 more parts I want you to review. 

Intro to Deliverance Ministry

Computers are capable of many impressive feats, but making you laugh is usually not one of them. Or could it be? This episode will talk about a custom-built machine learning algorithm that searches through text and writes jokes based on what it finds. 

The jokes are formulaic: they're all of the form "I like my X like I like my Y: Z" where X and Y are nouns, and Z is an adjective that can describe both X and Y. For (dumb) example, "I like my men like I like my coffee: steaming hot." The joke is funny when ZX and ZY are both very common phrases, but X and Y are rarely seen together. 

So, given a large enough corpus of text, the algorithm looks for triplets of words that fit this description and writes jokes based on them. Are the jokes funny? You be the judge...

http://homepages.inf.ed.ac.uk/s0894589/petrovic13unsupervised.pdf

Computers Try to Tell Jokes

A true classic topic in ML: Facial recognition is very high-dimensional, meaning that each picture can have millions of pixels, each of which can be a single feature. It's computationally expensive to deal with all these features, and invites overfitting problems. PCA (principal components analysis) is a classic dimensionality reduction tool that compresses these many dimensions into the few that contain the most variation in the data, and those principal components are often then fed into a classic ML algorithm like and SVM. 

One of the best thing about eigenfaces is the great example code that you can find in sklearn--you can distinguish pictures of world leaders yourself in just a few minutes!

http://scikit-learn.org/stable/auto_examples/applications/face_recognition.html

Facial Recognition with Eigenfaces

So, um... what about machine learning for audio applications?  In the course of starting this podcast, we've edited out a lot of "um"'s from our raw audio files.  It's gotten now to the point that, when we see the waveform in soundstudio, we can almost identify an "um" by eye.  Which makes it an interesting problem for machine learning--is there a way we can train an algorithm to recognize the "um" pattern, too?  This has become a little side project for Katie, which is very much still a work in progress.  We'll talk about what's been accomplished so far, some design choices Katie made in getting the project off the ground, and (of course) mistakes made and hopefully corrected.  We always say that the best way to learn something is by doing it, and this is our chance to try our own machine learning project instead of just telling you about what someone else did! 

Um Detector 1

The term spiritual warfare is used extensively in the church and in various contexts. A lot of the information that is shared is not necessarily wrong, rather can be vague and hard to apply.

This episode offers practical insights on how to wage spiritual warfare more effectively, from the perspective of two experience deliverance ministers and licensed Christian Counselors.

You will learn about:
driving our spirits (deliverance)
discerning spirits
resisting spirits
binding spirits

Episode #2:  Spiritual Warfare - A Deliverance Ministry Perspective

The reason that neural nets are taking over the world right now is because they can be efficiently trained with the backpropagation algorithm.  In short, backprop allows you to adjust the weights of the neural net based on how good of a job the neural net is doing at classifying training examples, thereby getting better and better at making predictions.  In this episode: we talk backpropagation, and how it makes it possible to train the neural nets we know and love.

Backpropagation

Wouldn’t it solve so many problems in the world if everyone just spoke the same language? Not so fast! 

Lingthusiasm is a brand-new podcast that’s enthusiastic about linguistics, hosted by Lauren Gawne of Superlinguo​ and Gretchen McCulloch of All Things Linguistic.

In this first episode of Lingthusiasm, ​Gretchen and Lauren discuss the “one language equals peace” fallacy, and whether speaking the same words means that people will necessarily agree with each other (spoiler: no). But the history of how people have tried is still really interesting, from constructed and symbolic communication like Blissymbols and emoji to the way astronauts communicate in the high stakes environment of the International Space Station.

For links to things mentioned during this episode visit http://lingthusiasm.com/post/154520059101/lingthusiasm-episode-1-speaking-a-single-language

Listen to bonus episodes, suggest future topics, and help keep the show ad-free by supporting us on Patreon: http://patreon.com/lingthusiasm

01: Speaking a single language won’t bring about world peace

Escape From The Tyranny Of Our Own Thoughts

Manly P. Hall - Escape From The Tyranny Of Our Own Thoughts

There are some pretty funny quotes of historical people complaining about kids back then doing linguistic things that now seem totally unremarkable. So let’s cut to the chase and celebrate linguistic innovation while it’s happening. 

In episode 7 of Lingthusiasm, your hosts Lauren Gawne and Gretchen McCulloch explore how far back we can trace complaints about the language of Kids These Days, why linguistic discrimination is harmful, and why “be like”, hyperbolic “literally”, and other modern innovations are actually signs of something awesome. 

We also announce a Patreon to keep the podcast sustainable! You can support us there to listen to a bonus episode about swearing and future monthly bonus content, and help decide the topics of future episodes. Even if you’re not sure about pledging, do check out the Patreon goals to see some of our future plans and our video to see a totally realistic and not at all staged cameo from our producer. Check out http://patreon.com/lingthusiasm 

For more information visit the show page: http://lingthusiasm.com/post/159796192161/lingthusiasm-episode-7-kids-these-days-arent

07: Kids these days aren’t ruining language

In this episode we briefly say what we're up to with the podcast.  No technical content, just a description of what each episode will look like, and why we're doing this.

00 - Intro to the podcast

https://www.semanticscholar.org/paper/A-Comparative-Study-of-Word-Embeddings-for-Reading-Dhingra-Liu/3ec37205c9201fc891ab51da200e361fdc34bfb3

01 - A Comparative Study of Word Embeddings for Reading Comprehension

https://www.semanticscholar.org/paper/Bidirectional-Attention-Flow-for-Machine-Seo-Kembhavi/007ab5528b3bd310a80d553cccad4b78dc496b02

02 - Bidirectional Attention Flow for Machine Comprehension

https://www.semanticscholar.org/paper/FastQA-A-Simple-and-Efficient-Neural-Architecture-Weissenborn-Wiese/7c1576b96a1e246d77b30f7b80cec63be96fa698

03 - FastQA: A Simple and Efficient Neural Architecture for Question Answering

An interview with Chris Dyer.

https://www.semanticscholar.org/paper/Recurrent-Neural-Network-Grammars-Dyer-Kuncoro/1594d954abc650bce2db445c52a76e49655efb0c

04 - Recurrent Neural Network Grammars, with Chris Dyer

https://www.semanticscholar.org/paper/Transition-Based-Dependency-Parsing-with-Stack-Lon-Dyer-Ballesteros/396b7932beac62a72288eaea047981cc9a21379a

05 - Transition-Based Dependency Parsing with Stack Long Short-Term Memory

https://www.semanticscholar.org/paper/Design-Challenges-for-Entity-Linking-Ling-Singh/aa2a7ac7bfa9a0201d4faddd4e7bb26299a5e0be

06 - Design Challenges for Entity Linking

https://www.semanticscholar.org/paper/Capturing-Semantic-Similarity-for-Entity-Linking-w-Francis-Landau-Durrett/1c9aca60f7ac5edcceb73d612806704a7d662643

07 - Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks

The recently-hyped paper that got "superhuman" performance on FAIR's CLEVR dataset.

https://arxiv.org/abs/1706.01427

20 - A simple neural network module for relational reasoning

How do we get from knowing words to making brand-new sentences out of them? In episode 9 of Lingthusiasm, your hosts Gretchen McCulloch and Lauren Gawne talk about how words form groups with other words: constituency. 

Once you start looking for it, constituency is everywhere: in ambiguous sentences like “time flies like an arrow; fruit flies like a banana”, in remixed films like “Of Oz The Wizard”, and even internet dog memes.

This month’s Patreon bonus was the backstory about the linguistics of the doggo meme and its connection to Australian slang, which grew out of this NPR article about doggo. You can get access to it and previous bonuses about swearing, teaching yourself linguistics, and explaining linguistics to employers by supporting Lingthusiasm on Patreon.

For more information, and links to things mentioned in this episode, visit the show page: http://lingthusiasm.com/post/161859273886/lingthusiasm-episode-9-the-bridge-between-words

09: The bridge between words and sentences - Constituency

Some linguists work with multiple languages, while others focus on just one. But for many people, learning a language after early childhood is the thing that first gets them curious about how language works in general and all the things in their native language(s) that they take for granted. 

In episode 10 of Lingthusiasm, your hosts Gretchen McCulloch and Lauren Gawne talk about how learning languages can feed into linguistics and vice versa. We also explore the power dynamics that affect learning languages, and the importance of learning about the rules of interaction as well as the rules of grammar.

This month’s Patreon bonus was about hypercorrection, where you try so hard to follow a linguistic rule that you end up overshooting. You can get access to it and previous bonuses about the doggo meme, swearing, teaching yourself linguistics, and explaining linguistics to employers by supporting Lingthusiasm on Patreon. http://patreon.com/lingthusiasm

Gretchen’s new recorder in this episode is thanks to the support of our patrons! 

For more information, and links to things mentioned in this episode, visit the show page: http://lingthusiasm.com/post/163226798571/lingthusiasm-episode-10-learning-languages

10: Learning languages linguistically

EMNLP 2017 paper by Nitish Gupta, Sameer Singh, and Dan Roth.

Nitish comes on to talk to us about his paper, which presents a new entity linking model that both unifies prior sources of information into a single neural model, and trains that model in a domain-agnostic way, so it can be transferred to new domains without much performance degradation.

https://www.semanticscholar.org/paper/Entity-Linking-via-Joint-Encoding-of-Types-Descrip-Gupta-Singh/a66b6a3ac0aa9af6c178c1d1a4a97fd14a882353

33 - Entity Linking via Joint Encoding of Types, Descriptions, and Context, with Nitish Gupta

These never before released lectures are considered by the thousands who have heard them to be profound and illuminating. Originally released by the Philosophical Research Society, Better Listen! is pleased to make available to the wider public a series of very special edition audio recordings from Manly P. Hall.

For the full download or other great tracks, visit
BetterListen.com or https://www.betterlisten.com/products/the-magnetic-fields-of-the-human-body-and-their-functions

Buy at BetterListen!

The Magnetic Fields of the Human Body and Their Functions with Manly P. Hall Preview 1

Paper by Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel R. Bowman.

Sam comes on to tell us about GLUE.  We talk about the motivation behind setting up a benchmark framework for natural language understanding, how the authors defined "NLU" and chose the tasks for this benchmark, a very nice diagnostic dataset that was constructed for GLUE, and what insight they gained from the experiments they've run so far.  We also have some musings about the utility of general-purpose sentence vectors, and about leaderboards.

https://www.semanticscholar.org/paper/GLUE%3A-A-Multi-Task-Benchmark-and-Analysis-Platform-Wang-Singh/a2054eff8b4efe0f1f53d88c08446f9492ae07c1

67 - GLUE: A Multi-Task Benchmark and Analysis Platform, with Sam Bowman

NAACL 2018 paper, by Rachel Rudinger, Aaron Steven White, and Benjamin Van Durme

Rachel comes on to the podcast, telling us about what factuality is (did an event happen?), what datasets exist for doing this task (a few; they made a new, bigger one), and how to build models to predict factuality (turns out a vanilla biLSTM does quite well).  Along the way, we have interesting discussions about how you decide what an "event" is, how you label factuality (whether something happened) on inherently uncertain text (like "I probably failed the test"), and how you might use a system that predicts factuality in some end task.

https://www.semanticscholar.org/paper/Neural-models-of-factuality-Rudinger-White/4d62a1e7819f9e3f8c837832c66659db5a6d9b37

68 - Neural models of factuality, with Rachel Rudinger

**The podcast is currently on hiatus. For more active NLP content, check out the Holistic Intelligence Podcast linked below.**

Welcome to the NLP highlights podcast, where we invite researchers to talk about their work in various areas in natural language processing. All views expressed belong to the hosts/guests, and do not represent their employers.

07 - Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks

NLP Highlights