Machine Learning has become an integral part of day to day life, particularly in the Digital World. Companies such as Google, Facebook, Netflix, and Amazon now use Machine Learning to continually update their services and algorithms. So far in this series, we have explored the basic principles, current applications and potential use of Machine Learning in SEO and Social. In this blog, we will look at some of the tools and applications that are actively being used in the market. The Machine Learning solutions below range from easy to use off-the-shelf solutions to code-heavy custom built solutions and platforms.

One question that always pops up in any machine learning problem: Which algorithm should I use? What do the algorithms do anyways? After briefly going over a typical machine learning process, we have a closer look at third step, i.e. building the model: What algorithms are out there? Which one should we use? One of Microsoft's Data Scientist, Brandon Rohrer, has written a nice three-part blog series on introducing data science with no jargon: Furthermore, there is one really neat cheat sheet created by Microsoft's Data Science team on when to use which algorithm: Finally, one last resource that I hihgly recommend: Top 10 data mining algorithms in plain English.

Not enough women are going into computer science. "I remember walking into one of the classes at Stanford and just deciding not to take the class because I was one of only three women there, and I just felt so intimidated," recalled Catherina Xu, one of the co-presidents for Women in Computer Science at Stanford University. Incidents like this are happening all across the country, and partly due to the lack of women in the field, there is now a shortage of computer science majors -- and it's going to get even worse. By 2024, the National Center for Women and Information Technology predicts that there will be 1.1 million computing-related job openings, and only 41% of those jobs will be filled. And get this: The percentage of women in the field has been declining since the 1980s.

Deep reinforcement learning (RL) can acquire complex behaviors from low-level inputs, such as images. However, real-world applications of such methods require generalizing to the vast variability of the real world. Deep networks are known to achieve remarkable generalization when provided with massive amounts of labeled data, but can we provide this breadth of experience to an RL agent, such as a robot? The robot might continuously learn as it explores the world around it, even while deployed. However, this learning requires access to a reward function, which is often hard to measure in real-world domains, where the reward could depend on, for example, unknown positions of objects or the emotional state of the user. Conversely, it is often quite practical to provide the agent with reward functions in a limited set of situations, such as when a human supervisor is present or in a controlled setting. Can we make use of this limited supervision, and still benefit from the breadth of experience an agent might collect on its own? In this paper, we formalize this problem as semisupervised reinforcement learning, where the reward function can only be evaluated in a set of "labeled" MDPs, and the agent must generalize its behavior to the wide range of states it might encounter in a set of "unlabeled" MDPs, by using experience from both settings. Our proposed method infers the task objective in the unlabeled MDPs through an algorithm that resembles inverse RL, using the agent's own prior experience in the labeled MDPs as a kind of demonstration of optimal behavior. We evaluate our method on challenging tasks that require control directly from images, and show that our approach can improve the generalization of a learned deep neural network policy by using experience for which no reward function is available. We also show that our method outperforms direct supervised learning of the reward.

Imagine that, in 20 or 30 years, a company creates the first artificially intelligent humanoid robot. She looks like a person, talks like a person, interacts like a person. If you were to meet Ava, you could relate to her even though you know she's a robot. Ava is a fully conscious, fully self-aware being: She communicates; she wants things; she improves herself. She is also, importantly, far more intelligent than her human creators.

Yogi Berra once said, "It's tough to make predictions, especially about the future." It is tough indeed, but enterprises that can make probabilistic predictions about customers, business processes, and operations will have an edge over enterprises that can't. These predictions don't have to be macroscopic to be consequential. Predictions about what a customer is likely to buy next. Predictions about marketing content that will resonate with a prospect.

We propose stochastic rank-$1$ bandits, a class of online learning problems where at each step a learning agent chooses a pair of row and column arms, and receives the product of their values as a reward. The main challenge of the problem is that the individual values of the row and column are unobserved. We assume that these values are stochastic and drawn independently. We propose a computationally-efficient algorithm for solving our problem, which we call Rank1Elim. We derive a $O((K + L) (1 / \Delta) \log n)$ upper bound on its $n$-step regret, where $K$ is the number of rows, $L$ is the number of columns, and $\Delta$ is the minimum of the row and column gaps; under the assumption that the mean row and column rewards are bounded away from zero. To the best of our knowledge, we present the first bandit algorithm that finds the maximum entry of a rank-$1$ matrix whose regret is linear in $K + L$, $1 / \Delta$, and $\log n$. We also derive a nearly matching lower bound. Finally, we evaluate Rank1Elim empirically on multiple problems. We observe that it leverages the structure of our problems and can learn near-optimal solutions even if our modeling assumptions are mildly violated.

This paper is a tutorial on Formal Concept Analysis (FCA) and its applications. FCA is an applied branch of Lattice Theory, a mathematical discipline which enables formalisation of concepts as basic units of human thinking and analysing data in the object-attribute form. Originated in early 80s, during the last three decades, it became a popular human-centred tool for knowledge representation and data analysis with numerous applications. Since the tutorial was specially prepared for RuSSIR 2014, the covered FCA topics include Information Retrieval with a focus on visualisation aspects, Machine Learning, Data Mining and Knowledge Discovery, Text Mining and several others.

The growth of data, the need for scalability and the complexity of models used in modern machine learning calls for distributed implementations. Yet, as of today, distributed machine learning frameworks have largely ignored the possibility of arbitrary (i.e., Byzantine) failures. In this paper, we study the robustness to Byzantine failures at the fundamental level of stochastic gradient descent (SGD), the heart of most machine learning algorithms. Assuming a set of $n$ workers, up to $f$ of them being Byzantine, we ask how robust can SGD be, without limiting the dimension, nor the size of the parameter space. We first show that no gradient descent update rule based on a linear combination of the vectors proposed by the workers (i.e, current approaches) tolerates a single Byzantine failure. We then formulate a resilience property of the update rule capturing the basic requirements to guarantee convergence despite $f$ Byzantine workers. We finally propose Krum, an update rule that satisfies the resilience property aforementioned. For a $d$-dimensional learning problem, the time complexity of Krum is $O(n^2 \cdot (d + \log n))$.

The mother of a special education student who drowned at his Chicago school's swimming pool earlier this year has filed a lawsuit claiming in part that her son was left unsupervised by a lifeguard who was using a computer in a nearby office. Rosario Gomez, an autistic 14-year-old student at Kennedy High School, was at the pool on Jan. 25 with a group of special education students, The Chicago Tribune reported. His mother, Yolanda Juarez, said that the district should have known that her son could not swim, and that his cognitive disabilities made it difficult for him to understand the dangers of the pool. The lawsuit alleges that Gomez was not paired with a buddy before entering the pool, and that the lifeguard supervising the group was in an office on the computer, The Chicago Tribune reported. The lawsuit claims Gomez "was allowed to struggle and drown while in the swimming pool without any intervention," and that he "was allowed to remain unnoticed at the bottom of the swimming pool" for a long enough period so that paramedics were unable to revive him, according to the report.