Over the last decade, Mexican drug cartels have been fighting each other--and corrupt police and military units--for control of the lucrative drug trade, plunging the country into chaos. Outsiders might think of Mexico as sunny and tequila-soaked, but beyond the beach resorts of Cancun and Mazatlan there hides a grimmer tale: levels of murder, rape, and kidnapping are hitting levels rarely seen outside hotspots in Africa, Asia, and South America. So grim the tale, when 43 college students went missing in Mexico's southern state of Guerrero in 2014, investigators found 129 other bodies in 60 fosas clandestinas (mass graves) before stumbling on badly burned remains in a mass grave they think might--possibly, maybe--contain what's left of the missing students. Mexico's attorney general says the local mayor conspired with the town's police force to abduct the students and turn them over to a local gang, who murdered them and burned the bodies, and dumped the charred corpses into a river. The situation is so bad that, after six decades of gains, the average life expectancy in Mexico has decreased, according to recent research.

Learning linear predictors with the logistic loss---both in stochastic and online settings---is a fundamental task in learning and statistics, with direct connections to classification and boosting. Existing "fast rates" for this setting exhibit exponential dependence on the predictor norm, and Hazan et al. (2014) showed that this is unfortunately unimprovable. Starting with the simple observation that the logistic loss is 1-mixable, we design a new efficient improper learning algorithm for online logistic regression that circumvents the aforementioned lower bound with a regret bound exhibiting a doubly-exponential improvement in dependence on the predictor norm. This provides a positive resolution to a variant of the COLT 2012 open problem of McMahan and Streeter (2012) when improper learning is allowed. This improvement is obtained both in the online setting and, with some extra work, in the batch statistical setting with high probability. We also show that the improved dependency on predictor norm is also near-optimal. Leveraging this improved dependency on the predictor norm yields the following applications: (a) we give algorithms for online bandit multiclass learning with the logistic loss with an $\tilde{O}(\sqrt{n})$ relative mistake bound across essentially all parameter ranges, thus providing a solution to the COLT 2009 open problem of Abernethy and Rakhlin (2009), and (b) we give an adaptive algorithm for online multiclass boosting with optimal sample complexity, thus partially resolving an open problem of Beygelzimer et al. (2015) and Jung et al. (2017). Finally, we give information-theoretic bounds on the optimal rates for improper logistic regression with general function classes, thereby characterizing the extent to which our improvement for linear classes extends to other parameteric and even nonparametric settings.

Recent advances in policy gradient methods and deep learning have demonstrated their applicability for complex reinforcement learning problems. However, the variance of the performance gradient estimates obtained from the simulation is often excessive, leading to poor sample efficiency. In this paper, we apply the stochastic variance reduced gradient descent (SVRG) to model-free policy gradient to significantly improve the sample-efficiency. The SVRG estimation is incorporated into a trust-region Newton conjugate gradient framework for the policy optimization. On several Mujoco tasks, our method achieves significantly better performance compared to the state-of-the-art model-free policy gradient methods in robotic continuous control such as trust region policy optimization (TRPO)

World Bank hosted its poverty prediction competition on the competition hosting website drivendata.org. The link to the competition is here. We decided to try out our Machine Learning skills on this dataset. Most regular work in ParallelDots is around three themes: Visual Analytics on images and videos, Healthcare AI and NLP, all three of which are solved using Deep Learning techniques. This competition was a chance to try out something new and build our internal codebase to handle tabular datasets like what we had in the competition.

Modern artificial intelligence is way beyond playing chess; it has mastered Go and kicks butt in Dota 2, among other games. What started as a test-lab monkey has evolved into something akin to a prodigy child. Artificial intelligence, or AI, may still have to be fed information, but once it has gathered enough, it can come up with results that mimic the original data. First came the static images -- AI managed to create perfectly convincing images of people who have never existed. Then it showed it was perfectly capable of mimicking different seasons.

What Tomorrow's Business Leaders Need to Know About Machine Learning Sometimes I write a blog just to formulate and organize a point of view, and I think it's time that I pull together the bounty of excellent information about Machine Learning. This is a topic with which business leaders must become comfortable, especially tomorrow's business leaders (tip for my next semester University of San Francisco business students!). Machine learning is a key capability that will help organizations drive optimization and monetization opportunities, and there have been some recent developments that will place basic machine learning capabilities into the hands of the lines of business. By the way, there is an absolute wealth of freely-available material on machine learning, so I've included a sources section at the end of this blog for folks who want more details on machine learning. Time to dive into the world of machine learning!

The randomized experiment is an important tool for inferring the causal impact of an intervention. The recent literature on statistical learning methods for heterogeneous treatment effects demonstrates the utility of estimating the marginal conditional average treatment effect (MCATE), i.e., the average treatment effect for a subpopulation of respondents who share a particular subset of covariates. However, each proposed method makes its own set of restrictive assumptions about the intervention's effects, the underlying data generating processes, and which subpopulations (MCATEs) to explicitly estimate. Moreover, the majority of the literature provides no mechanism to identify which subpopulations are the most affected--beyond manual inspection--and provides little guarantee on the correctness of the identified subpopulations. Therefore, we propose Treatment Effect Subset Scan (TESS), a new method for discovering which subpopulation in a randomized experiment is most significantly affected by a treatment. We frame this challenge as a pattern detection problem where we maximize a nonparametric scan statistic (measurement of distributional divergence) over subpopulations, while being parsimonious in which specific subpopulations to evaluate. Furthermore, we identify the subpopulation which experiences the largest distributional change as a result of the intervention, while making minimal assumptions about the intervention's effects or the underlying data generating process. In addition to the algorithm, we demonstrate that the asymptotic Type I and II error can be controlled, and provide sufficient conditions for detection consistency---i.e., exact identification of the affected subpopulation. Finally, we validate the efficacy of the method by discovering heterogeneous treatment effects in simulations and in real-world data from a well-known program evaluation study.

We need machine learning for tasks that are too complex for humans to code directly, i.e. tasks that are so complex that it is impractical, if not impossible, for us to work out all of the nuances and code for them explicitly. So instead, we provide a machine learning algorithm with a large amount of data and let it explore and search for a model that will work out what the programmers have set out to achieve. Let's look at these two examples: Then comes the Machine Learning Approach: instead of writing a program by hand for each specific task, we collect lots of examples that specify the correct output for a given input. A machine learning algorithm then takes these examples and produces a program that does the job. The program produced by the learning algorithm may look very different from a typical hand-written program -- it may contain millions of numbers. If we do it right, the program works for new cases, as well as the ones we trained it on. If the data changes, the program can change too by training from the new data. You should note that conducting massive amounts of computation is now cheaper than paying someone to write a task-specific program.

In 1900, 30 million people in the United States were farmers. By 1990 that number had fallen to under 3 million even as the population more than tripled. So, in a matter of speaking, 90% of American agriculture workers lost their jobs, mostly due to automation. Yet somehow, the 20th century was still seen as an era of unprecedented prosperity. In the decades to come, we are likely to see similar shifts.

There are tons of resources to help you learn the different aspects of R, and as a beginner this can be overwhelming. It's also a dynamic language and rapidly changing, so it's important to keep up with the latest tools and technologies. That's why R-bloggers and DataCamp have worked together to bring you a learning path for R. Each section points you to relevant resources and tools to get you started and keep you engaged to continue learning. Just like R, this learning path is a dynamic resource.