Linear Digressions

If one machine learning model is good, are two models better? In a lot of cases, the answer is yes. If you build many ok models, and then bring them all together and use them in combination to make your final predictions, you've just created an ensemble model. It feels a little bit like cheating, like you just got something for nothing, but the results don't like: algorithms like Random Forests and Gradient Boosting Trees (two types of ensemble algorithms) are some of the strongest out-of-the-box algorithms for classic supervised classification problems. What makes a Random Forest random, and what does it mean to gradient boost a tree? Have a listen and find out.

As anyone who's encountered a badly translated text could tell you, not all translations are created equal. Some translations are smooth, fluent and sound like a poet wrote them; some are jerky, non-grammatical and awkward. When a machine is doing the translating, it's awfully easy to end up with a robotic-sounding text; as the state of the art in machine translation improves, though, a natural question to ask is: according to what measure? How do we quantify a "good" translation? Enter the BLEU score, which is the standard metric for quantifying the quality of a machine translation. BLEU rewards translations that have large overlap with human translations of sentences, with some extra heuristics thrown in to guard against weird pathologies (like full sentences getting translated as one word, redundancies, and repetition). Nowadays, if there's a machine translation being evaluated or a new state-of-the-art system (like the Google neural machine translation we've discussed on this podcast before), chances are that there's a BLEU score going into that assessment.

Take Google-size data, the flexibility of a neural net, and all (well, most) of the languages of the world, and what you end up with is a pile of surprises. This episode is about some interesting features of Google's new neural machine translation system, namely that with minimal tweaking, it can accommodate many different languages in a single neural net, that it can do a half-decent job of translating between language pairs it's never been explicitly trained on, and that it seems to have its own internal representation of concepts that's independent of the language those concepts are being represented in. Intrigued? You should be...

Recently, Google swapped out the backend for Google Translate, moving from a statistical phrase-based method to a recurrent neural network. This marks a big change in methodology: the tried-and-true statistical translation methods that have been in use for decades are giving way to a neural net that, across the board, appears to be giving more fluent and natural-sounding translations. This episode recaps statistical phrase-based methods, digs into the RNN architecture a little bit, and recaps the impressive results that is making us all sound a little better in our non-native languages.

Today we are delighted to bring you an interview with Matt Might, computer scientist and medical researcher extraordinaire and architect of President Obama's Precision Medicine Initiative. As the Obama Administration winds down, we're talking with Matt about the goals and accomplishments of precision medicine (and related projects like the Cancer Moonshot) and what he foresees as the future marriage of data and medicine. Many thanks to Matt, our friends over at Partially Derivative (hi, Jonathon!) and the White House for arranging this opportunity to chat. Enjoy!

We have the pleasure of bringing you a very special crossover episode this week: our friends at Partially Derivative (another great podcast about data science, you should check it out) recently interviewed White House Chief Data Scientist DJ Patil. We think DJ's message about the importance and impact of data science is worth spreading, so it's our pleasure to bring it to you today. A huge thanks to Jonathon Morgan and Partially Derivative for sharing this interview with us--enjoy! Relevant links: http://partiallyderivative.com/podcast/2016/12/13/dj-patil

Competing in a machine learning competition on Kaggle is a kind of rite of passage for data scientists. Losing unexpectedly at the very end of the contest is also something that a lot of us have experienced. It's not just bad luck: a very specific combination of overfitting on popular competitions can take someone who is in the top few spots in the final days of a contest and bump them down hundreds of slots in the final tally.

Imagine there's an important decision to be made about someone, like a bank deciding whether to extend a loan, or a school deciding to admit a student--unfortunately, we're all too aware that discrimination can sneak into these situations (even when everyone is acting with the best of intentions!). Now, these decisions are often made with the assistance of machine learning and statistical models, but unfortunately these algorithms pick up on the discrimination in the world (it sneaks in through the data, which can capture inequities, which the algorithms then learn) and reproduce it. This podcast covers some of the most common ways we can try to minimize discrimination, and why none of those ways is perfect at fixing the problem. Then we'll get to a new idea called "equality of opportunity," which came out of Google recently and takes a pretty practical and well-aimed approach to machine learning bias.

This week, we're doing a crash course in recurrent neural networks--what the structural pieces are that make a neural net recurrent, how that structure helps RNNs solve certain time series problems, and the importance of forgetfulness in RNNs. Relevant links: http://colah.github.io/posts/2015-08-Understanding-LSTMs/

Want another reason to be paranoid when using the free coffee shop wifi? Allow us to introduce WindTalker, a system that cleverly combines a dose of signal processing with a dash of machine learning to (potentially) steal the PIN from your phone transactions without ever having physical access to your phone. This episode has it all, folks--channel state information, ICMP echo requests, low-pass filtering, PCA, dynamic time warps, and the PIN for your phone.

Cryptography used to be the domain of information theorists and spies. There's a new player now: neural networks. Given the task of communicating securely, neural networks are inventing new encryption methods that, as best we can tell, are unlike anything humans have ever seen before. Relevant links: http://arstechnica.co.uk/information-technology/2016/10/google-ai-neural-network-cryptography/ https://arxiv.org/pdf/1610.06918v1.pdf

In 1997, Deep Blue was the IBM algorithm/computer that did what no one, at the time, though possible: it beat the world's best chess player. It turns out, though, that one of the most important moves in the matchup, where Deep Blue psyched out its opponent with a weird move, might not have been so inspired after all. It might have been nothing more than a bug in the program, and it changed computer science history. Relevant links: https://www.wired.com/2012/09/deep-blue-computer-bug/

If you're a data scientist, there's a good chance you're used to working with a lot of data. But there's a lot of data, and then there's Google-scale amounts of data. Keeping all that data organized is a Google-sized task, and as it happens, they've built a system for that organizational challenge. This episode is all about that system, called Goods, and in particular we'll dig into some of the details of what makes this so tough. Relevant links: http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45390.pdf

A few months ago, Katie started on a project for the Vice President's Cancer Moonshot surrounding how data can be used to better fight cancer. The project is all wrapped up now, so we wanted to tell you about how that work went and what changes to cancer data policy were suggested to the Vice President. See lineardigressions.com for links to the reports discussed on this episode.

Sick of the presidential election yet? We are too, but there's still almost a month to go, so let's just embrace it together. This week, we'll talk about one of the presidential polls, which has been kind of an outlier for quite a while. This week, the NY Times took a closer look at this poll, and was able to figure out the reason it's such an outlier. It all goes back to a 19-year-old African American man, living in Illinois, who really likes Donald Trump... Relevant Links: http://www.nytimes.com/2016/10/13/upshot/how-one-19-year-old-illinois-man-is-distorting-national-polling-averages.html followup article from LA Times, released after recording: http://www.latimes.com/politics/la-na-pol-daybreak-poll-questions-20161013-snap-story.html

What does it mean to steal a model? It means someone (the thief, presumably) can re-create the predictions of the model without having access to the algorithm itself, or the training data. Sound far-fetched? It isn't. If that person can ask for predictions from the model, and he (or she) asks just the right questions, the model can be reverse-engineered right out from under you. Relevant links: https://www.usenix.org/system/files/conference/usenixsecurity16/sec16_paper_tramer.pdf

Lots of data is usually seen as a good thing. And it is a good thing--except when it's not. In a lot of fields, a problem arises when you have many, many features, especially if there's a somewhat smaller number of cases to learn from; supervised machine learning algorithms break, or learn spurious or un-interpretable patterns. What to do? Regularization can be one of your best friends here--it's a method that penalizes overly complex models, which keeps the dimensionality of your model under control.

You might sometimes find that it's hard to get started doing something, but once you're going, it gets easier. Turns out machine learning algorithms, and especially recommendation engines, feel the same way. The more they "know" about a user, like what movies they watch and how they rate them, the better they do at suggesting new movies, which is great until you realize that you have to start somewhere. The "cold start" problem will be our focus in this episode, both the heuristic solutions that help deal with it and a bit of realism about the importance of skepticism when someone claims a great solution to cold starts. Relevant links: http://repository.upenn.edu/cgi/viewcontent.cgi?article=1141&context=cis_papers

If you work in tech, software or data science, there's an excellent chance you use tools that are built upon open source software. This is software that's built and distributed not for a profit, but because everyone benefits when we work together and share tools. Tim Head of scikit-optimize chats with us further about what it's like to maintain an open source library, how to get involved in open source, and why people like him need people like you to make it all work.

We're excited to welcome a guest, Tim Head, who is one of the maintainers of the scikit-optimize package. With all the talk about optimization lately, it felt appropriate to get in a few words with someone who's out there making it happen for python. Relevant links: https://scikit-optimize.github.io/ http://www.wildtreetech.com/

It's a funny thing to realize, but data science modeling is usually about either explainability, interpretation and understanding, or it's about predictive accuracy. But usually not both--optimizing for one tends to compromise the other. Leo Breiman was one of the titans of both kinds of modeling, a statistician who helped bring machine learning into statistics and vice versa. In this episode, we unpack one of his seminal papers from 2001, when machine learning was just beginning to take root, and talk about how he made clear what machine learning could do for statistics and why it's so important. Relevant links: http://www.math.snu.ac.kr/~hichoi/machinelearning/(Breiman)%20Statistical%20Modeling--The%20Two%20Cultures.pdf

You've got an optimization problem to solve, and a less-than-forever amount of time in which to solve it. What do? Use a heuristic optimization algorithm, like a hill climber or simulated annealing--we cover both in this episode! Relevant link: http://www.lizsander.com/programming/2015/08/04/Heuristic-Search-Algorithms.html

If modeling is about predicting the unknown, optimization tries to answer the question of what to do, what decision to make, to get the best results out of a given situation. Sometimes that's straightforward, but sometimes... not so much. What makes an optimization problem easy or hard, and what are some of the methods for finding optimal solutions to problems? Glad you asked! May we recommend our latest podcast episode to you?

Ok, this episode is only sort of about DEADLY RADIOACTIVE GAS. It's mostly about multilevel modeling, which is a way of building models with data that has distinct, related subgroups within it. What are multilevel models used for? Elections (we can't get enough of 'em these days), understanding the effect that a good teacher can have on their students, and DEADLY RADIOACTIVE GAS. Relevant links: http://www.stat.columbia.edu/~gelman/research/published/multi2.pdf

Continuing the discussion of how polls do (and sometimes don't) tell us what to expect in upcoming elections--let's take a concrete example from the recent past, shall we? The Brexit referendum was, by and large, expected to shake out for "remain", but when the votes were counted, "leave" came out ahead. Everyone was shocked (SHOCKED!) but maybe the polls weren't as wrong as the pundits like to claim. Relevant links: http://www.slate.com/articles/news_and_politics/moneybox/2016/07/why_political_betting_markets_are_failing.html http://andrewgelman.com/2016/06/24/brexit-polling-what-went-wrong/

Not sure if you heard, but there's an election going on right now. Polls, surveys, and projections about, as far as the eye can see. How to make sense of it all? How are the projections made? Which are some good ones to follow? We'll be your trusty guides through a crash course in election forecasting. Relevant links: http://www.wired.com/2016/06/civis-election-polling-clinton-sanders-trump/ http://election.princeton.edu/ http://projects.fivethirtyeight.com/2016-election-forecast/ http://www.nytimes.com/interactive/2016/upshot/presidential-polls-forecast.html?rref=collection%2Fsectioncollection%2Fupshot&action=click&contentCollection=upshot&region=rank&module=package&version=highlights&contentPlacement=5&pgtype=sectionfront

Genomics data is some of the biggest #bigdata, and doing machine learning on it is unlocking new ways of thinking about evolution, genomic diseases like cancer, and what really makes each of us different for everyone else. This episode touches on some of the things that make machine learning on genomics data so challenging, and the algorithms designed to do it anyway.

Hot enough for you? Climate models suggest that it's only going to get warmer in the coming years. This episode unpacks those models, so you understand how they work. A lot of the episodes we do are about fun studies we hear about, like "if you're interested, this is kinda cool"--this episode is much more important than that. Understanding these models, and taking action on them where appropriate, will have huge implications in the years to come. Relevant links: https://climatesight.org/

Last week’s episode on artificial intelligence gets a huge payoff this week—we’ll explore a wonderful couple of papers about all the ways that artificial intelligence can go wrong. Malevolent actors? You bet. Collateral damage? Of course. Reward hacking? Naturally! It’s fun to think about, and the discussion starting now will have reverberations for decades to come. https://www.technologyreview.com/s/601519/how-to-create-a-malevolent-artificial-intelligence/ http://arxiv.org/abs/1605.02817 https://arxiv.org/abs/1606.06565

There’s a ton of excitement about reinforcement learning, a form of semi-supervised machine learning that underpins a lot of today’s cutting-edge artificial intelligence algorithms. Here’s a crash course in the algorithmic machinery behind AlphaGo, and self-driving cars, and major logistical optimization projects—and the robots that, tomorrow, will clean our houses and (hopefully) not take over the world…

Apple wants to study iPhone users' activities and use it to improve performance. Google collects data on what people are doing online to try to improve their Chrome browser. Do you like the idea of this data being collected? Maybe not, if it's being collected on you--but you probably also realize that there is some benefit to be had from the improved iPhones and web browsers. Differential privacy is a set of policies that walks the line between individual privacy and better data, including even some old-school tricks that scientists use to get people to answer embarrassing questions honestly. Relevant links: http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42852.pdf

Something a little different in this episode--we'll be talking about the technical plumbing that gets our podcast from our brains to your ears. As it turns out, it's a multi-step bucket brigade process of RSS feeds, links to downloads, and lots of hand-waving when it comes to trying to figure out how many of you (listeners) are out there.

Machine learning on imbalanced classes: surprisingly tricky. Many (most?) algorithms tend to just assign the majority class label to all the data and call it a day. SMOTE is an algorithm for manufacturing new minority class examples for yourself, to help your algorithm better identify them in the wild. Relevant links: https://www.jair.org/media/953/live-953-2037-jair.pdf

Conjoint analysis is like AB tester, but more bigger more better: instead of testing one or two things, you can test potentially dozens of options. Where might you use something like this? Well, if you wanted to design an entire hotel chain completely from scratch, and to do it in a data-driven way. You'll never look at Courtyard by Marriott the same way again. Relevant link: https://marketing.wharton.upenn.edu/files/?whdmsaction=public:main.file&fileID=466

This episode is for all you (us) traffic nerds--we're talking about the hidden structure underlying traffic on-ramp metering systems. These systems slow down the flow of traffic onto highways so that the highways don't get overloaded with cars and clog up. If you're someone who listens to podcasts while commuting, and especially if your area has on-ramp metering, you'll never look at highway access control the same way again (yeah, we know this is super nerdy; it's also super awesome). Relevant links: http://its.berkeley.edu/sites/default/files/publications/UCB/99/PWP/UCB-ITS-PWP-99-19.pdf http://www.its.uci.edu/~lchu/ramp/Final_report_mou3013.pdf

One tricky thing about working with time series data, like the audio data in our "um" detector (remember that? because we barely do...), is that sometimes events look really similar but one is a little bit stretched and squeezed relative to the other. Besides having an amazing name, the dynamic time warp is a handy algorithm for aligning two time series sequences that are close in shape, but don't quite line up out of the box. Relevant link: http://www.aaai.org/Papers/Workshops/1994/WS-94-03/WS94-03-031.pdf

It's storytime this week--the story, from beginning to end, of how Katie designed and built the main project for Udacity's Intro to Machine Learning class, when she was developing the course. The project was to use email and financial data to hunt for signatures of fraud at Enron, one of the biggest cases of corporate fraud in history; that description makes the project sound pretty clean but getting the data into the right shape, and even doing some dataset merging (that hadn't ever been done before), made this project much more interesting to design than it might appear. Here's the story of what a data analysis like this looks like...from the inside.

Data science and is often mentioned in the same breath as big data. But how big is big data? And who has the biggest big data? CERN? Youtube? ... Something (or someone) else? Relevant link: http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1002195

Supervised machine learning assumes that the features and labels used for building a classifier are isolated from each other--basically, that you can't cheat by peeking. Turns out this can be easier said than done. In this episode, we'll talk about the many (and diverse!) cases where label information contaminates features, ruining data science competitions along the way. Relevant links: https://www.researchgate.net/profile/Claudia_Perlich/publication/221653692_Leakage_in_data_mining_Formulation_detection_and_avoidance/links/54418bb80cf2a6a049a5a0ca.pdf

Machine learning algorithms can be black boxes--inputs go in, outputs come out, and what happens in the middle is anybody's guess. But understanding how a model arrives at an answer is critical for interpreting the model, and for knowing if it's doing something reasonable (one could even say... trustworthy). We'll talk about a new algorithm called LIME that seeks to make any model more understandable and interpretable. Relevant Links: http://arxiv.org/abs/1602.04938 https://github.com/marcotcr/lime/tree/master/lime

We've got updates for you about topics from past shows! First, the political science scandal of the year 2015 has a new chapter, we'll remind you about the original story and then dive into what has happened since. Then, we've got an update on AlphaGo, and his/her/its much-anticipated match against the human champion of the game Go. Relevant Links: https://soundcloud.com/linear-digressions/electoral-insights-part-2 https://soundcloud.com/linear-digressions/go-1 http://www.sciencemag.org/news/2016/04/talking-people-about-gay-and-transgender-issues-can-change-their-prejudices http://science.sciencemag.org/content/sci/352/6282/220.full.pdf http://qz.com/639952/googles-ai-won-the-game-go-by-defying-millennia-of-basic-human-instinct/ http://www.wired.com/2016/03/two-moves-alphago-lee-sedol-redefined-future/ http://www.wired.com/2016/03/sadness-beauty-watching-googles-ai-play-go/

Simpson's paradox is the data science equivalent of looking through one eye and seeing a very clear trend, and then looking through the other eye and seeing the very clear opposite trend. In one case, you see a trend one way in a group, but then breaking the group into subgroups gives the exact opposite trend. Confused? Scratching your head? Welcome to the tricky world of ecological inference. Relevant links: https://gking.harvard.edu/files/gking/files/part1.pdf http://blog.revolutionanalytics.com/2013/07/a-great-example-of-simpsons-paradox.html

Sometimes when we say an algorithm discriminates, we mean it can tell the difference between two types of items. But in this episode, we'll talk about another, more troublesome side to discrimination: algorithms can be... racist? Sexist? Ageist? Yes to all of the above. It's an important thing to be aware of, especially when doing people-centered data science. We'll discuss how and why this happens, and what solutions are out there (or not). Relevant Links: http://www.nytimes.com/2015/07/10/upshot/when-algorithms-discriminate.html http://techcrunch.com/2015/08/02/machine-learning-and-human-bias-an-uneasy-pair/ http://www.sciencefriday.com/segments/why-machines-discriminate-and-how-to-fix-them/ https://medium.com/@geomblog/when-an-algorithm-isn-t-2b9fe01b9bb5#.auxqi5srz

This episode started out as a discussion of recommendation engines, like Netflix uses to suggest movies. There's still a lot of that in here. But a related topic, which is both interesting and important, is how to keep data private in the era of large-scale recommendation engines--what mistakes have been made surrounding supposedly anonymized data, how data ends up de-anonymized, and why it matters for you. Relevant links: http://www.netflixprize.com/ http://bits.blogs.nytimes.com/2010/03/12/netflix-cancels-contest-plans-and-settles-suit/?_r=0 http://arxiv.org/PS_cache/cs/pdf/0610/0610105v2.pdf

One neural net is creating counterfeit bills and passing them off to a second neural net, which is trying to distinguish the real money from the fakes. Result: two neural nets that are better than either one would have been without the competition. Relevant links: http://arxiv.org/pdf/1406.2661v1.pdf http://arxiv.org/pdf/1412.6572v3.pdf http://soumith.ch/eyescream/

In this episode, we're taking many episodes' worth of insights and unpacking an extremely complex and important question--in what ways are we winning the fight against cancer, where might that fight go in the coming decade, and how do we know when we're making progress? No matter how tricky you might think this problem is to solve, the fact is, once you get in there trying to solve it, it's even trickier than you thought.

Hey, sick of the election yet? Fear not, there are algorithms that can automagically generate political-ish speech so that we never need to be without an endless supply of Congressional speeches and Donald Trump twitticisms! Relevant links: http://arxiv.org/pdf/1601.03313v2.pdf http://qz.com/631497/mit-built-a-donald-trump-ai-twitter-bot-that-sounds-scarily-like-him/ https://twitter.com/deepdrumpf

Multi-armed bandits: how to take your randomized experiment and make it harder better faster stronger. Basically, a multi-armed bandit experiment allows you to optimize for both learning and making use of your knowledge at the same time. It's what the pros (like Google Analytics) use, and it's got a great name, so... winner! Relevant link: https://support.google.com/analytics/answer/2844870?hl=en

"People with a family history of heart disease are more likely to eat healthy foods, and have a high incidence of heart attacks." Did the healthy food cause the heart attacks? Probably not. But establishing causal links is extremely tricky, and extremely important to get right if you're trying to help students, test new medicines, or just optimize a website. In this episode, we'll unpack randomized experiments, like AB tests, and maybe you'll be smarter as a result. Will you be smarter BECAUSE of this episode? Well, tough to say for sure... Relevant link: http://tylervigen.com/spurious-correlations

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.