Ask an MLer: P values and Permutation Tests

Ask an MLer: P values and Permutation Tests

Ask an MLer: Classification and Hypothesis Testing Often scientists want to use machine learning as an analytical tool to answer a question about a data set. However, transitioning from classification accuracy to a hypothesis test can be tricky. Today we answer a question from a neuroscientist on the subject. How...
OHBM Annual Meeting 2017

OHBM Annual Meeting 2017

This year I presented a poster at the Organization for Human Brain Mapping (OHBM) annual meeting in Vancouver. I had a nice time and saw some interesting stuff, which I’d like to share with you all. To keep it brief, I’m going to focus on the highlights. Overall Themes Reproducibility....
How can we show that deep learning and the brain are related? Integration of Deep Learning and Neuroscience Final Post

How can we show that deep learning and the brain are related? Integration of Deep Learning and Neuroscience Final Post

As we’ve discussed in our previous posts, this paper on the integration of deep learning and neuroscience has been highly speculative. The authors have listed deep learning-inspired hypotheses about the brain and discussed how the brain may be consistent with those hypotheses. The concluding portion of the paper discusses potential...
Latest entries
Brainy Breakdown: 7/21/17

Brainy Breakdown: 7/21/17

Explaining ML with food analogies. Basic but fun. The battle for fMRI cluster sizes rages on. Confirmed: Ravens are really smart! AI and Super Smash Bros, Melee Want to inspire a young potential scientist? Become their penpal! What are humans, anyway?  
Brainy Breakdown: 7/14/17

Brainy Breakdown: 7/14/17

Thanks to a tremendous effort, we’re closer than ever to understanding the drosophila brain Can you see evidence of poverty in the brain? This article is pretty upsetting – we’ll need to check out the actual papers soon. Scientific racism through the ages This includes discussion of a paper we’ve reviewed here. The value of an...
Brainy Breakdown: 7/9/2017

Brainy Breakdown: 7/9/2017

We should care where the money for our research comes from. And what our research is used for! Also: we should care about where it’s published. Support Open Access! People share sexism they’ve experienced in academia. Note: you will be depressed by what you read On a lighter note, Alona is helping to organize a...
Brainy Breakdown: 6/30/2017

Brainy Breakdown: 6/30/2017

This past week was the OHBM annual meeting – stay tuned for a trip report! Excellent piece at the NYT on what actually makes AI dangerous. I don’t know enough about economics to judge the proposed solutions, though. Wired thinks we should all just calm down about superhuman AI Related, AI-generated inspirational memes. A favorite....
Brainy Breakdown: 6/23/2017

Brainy Breakdown: 6/23/2017

Confrontational title aside, this is a cool interview with someone who’s done interesting stuff with google search data. Considering ordering the book. Do you love deep learning and hate waiting in line at the airport? Here’s a contest for you! This week I was introduced to the concept of robot sumo wrestling. (Note: the sound...
Brainy Breakdown: 6/16/2107

Brainy Breakdown: 6/16/2107

How can we get more women into AI? While recruitment is very important, retention is equally important, if not more so.   An adversarial review of GANs for NLP. #NotAllDeepLearners?  And a response from Fernando Pereira.  Another response from Yann LeCun, and then (of course) a response to Yann’s response.     Will “skilled in...
Brainy Breakdown: 6/2/2017

Brainy Breakdown: 6/2/2017

Want to be perceived as a good scientist by randos? Try looking less attractive. The Department of Labor is investigating Google’s gender wage gap. I guess not all biases were busted. Related: https://twitter.com/tef_ebooks/status/868867367639670784 Can scientists write better? On the importance of sharing code in neuroscience. I’m 100% on board with this. Can a computer know...
Brainy Breakdown: 5/26/2017

Brainy Breakdown: 5/26/2017

DeepMind’s AlphaGo achieves superhuman performance, becoming “the god of the Go game” A surprisingly reasonable discussion about the singularity. Note that they don’t use the word. Related: AI can be overhyped. On the positive side, Sebastian Thrun on the power of AI. Privacy concerns who? Google takes on AutoML to try to automate the myriad...
Paper Review: An Integration of Deep Learning and Neuroscience, Part 3

Paper Review: An Integration of Deep Learning and Neuroscience, Part 3

In our previous post we talked about how the brain might optimize cost functions. Now we’ll explore how cost functions may be generated, represented, and change over time in the brain. Marblestone et al outline several ways that cost functions could be generated.  In particular, they talk about specialized circuitry for comparing the predicted output...