Over the last year, we have heard leading minds warning of the dangers of artificial intelligence (AI). We've since seen a robot that can take a beating and get back up unfazed, and borne witness to a computer that can best a Go master at his extremely complex game. Find and replace "Google" with a nefarious "Skynet" clone and you have yourself the start of a film franchise. Before it inevitably takes over the world, what impact will AI and robotics actually have in day-to-day business -- and who might benefit? When Arnold Schwarzenegger's T-800 character drove his Chevrolet Nova through the police station window in response to the "unhelpful" duty officer, I suspect not many people thought he was after his job.

Very interesting... this book is the volume # 31 in the series "Handbook of Statistics" edited by C.R. Rao and V. Govindaraju. The series started well before I completed my PhD in 1993, and obviously, they believe machine learning is a sub-domain of statistics. This might be the most comprehensive attempt at producing an exhaustive resource for statistician, with more than 10,000 pages to date. Second interesting fact: I purchased the book on Amazon for less than 100, last week. Now when I check it, it is listed at 184.35.

Just when you thought you'd got your head around the whole Machine Learning thing…BAMN! There's a new tech buzzword in town rearing up to take it's place. And while it may seem like just another Silicon Valley buzzword that all the new startups will claim to be using, deep learning is actually already being used to make some really astounding advances. We caught up with deep learning expert, Andrew Ng, and asked him to explain what deep learning is and how we should expect to see it change the world in 2016. Deep learning is a subset of machine learning that essentially refers to trying to map neural networks (the same stuff that makes your brain work).

A few years ago, Magnus Egerstedt was walking through the swarm robotics laboratory at the Georgia Institute of Technology, where he is associate director of research, feeling proud of the research spearheaded there, when a disturbing thought crossed his mind. "I began thinking about the robotics laboratories where people are doing things that matter. There's not even ten of them globally," Egerstedt says. "That's weird, because so many people are working on swarm robotics, but it takes money and people to drive research that matters. He immediately envisioned a way to give robotics researchers who aren't with those top labs access to top-lab capabilities. And he knew students at all levels, grade school to graduate school, could benefit as well. "I used as a model the Large Hadron Collider," Egerstedt says. "Physicists realized large particle colliders were too expensive to build separately, so they share.

It is obvious that we are living in a data deluge era, evidenced by the phenomenon that enormous amount of data have been being continually generated at unprecedented and ever increasing scales. Large-scale data sets are collected and studied in numerous domains, from engineering sciences to social networks, commerce, biomolecular research, and security [1]. Particularly, digital data, generated from a variety of digital devices, are growing at astonishing rates. According to [2], in 2011, digital information has grown nine times in volume in just 5 years and its amount in the world will reach 35 trillion gigabytes by 2020 [3]. Therefore, the term "Big Data" was coined to capture the profound meaning of this data explosion trend. To clarify what the big data refers to, several good surveys have been presented recently and each of them views the big data from different perspectives, including challenges and opportunities [4], background and research status [5], and analytics platforms [6].

You've probably been there: Something stressful is happening in your life, and you're feeling more anxious than usual. You'd love to talk to someone about it, but you don't know who to turn to. Therapy is one option, but A) it can be crazy expensive, and B) you don't want to be that person who has to see a shrink. Turns out, there is no that person. Roughly one in every five Americans, or about 43 million people, suffers from mental illness, according to the National Institute of Mental Health.

This paper characterizes the conditional distribution properties of the finite sample ridge regression estimator and uses that result to evaluate total regression and generalization errors that incorporate the inaccuracies committed at the time of parameter estimation. The paper provides explicit formulas for those errors. Unlike other classical references in this setup, our results take place in a fully singular setup that does not assume the existence of a solution for the non-regularized regression problem. In exchange, we invoke a conditional homoscedasticity hypothesis on the regularized regression residuals that is crucial in our developments.

We propose a scheme for recycling Gaussian random vectors into structured matrices to approximate various kernel functions in sublin-ear time via random embeddings. Our framework includes the Fastfood construction of Le et al. (2013) as a special case, but also extends to Circulant, Toeplitz and Hankel matrices, and the broader family of structured matrices that are characterized by the concept of low-displacement rank. We introduce notions of coherence and graph-theoretic structural constants that control the approximation quality, and prove unbiasedness and low-variance properties of random feature maps that arise within our framework. For the case of low-displacement matrices, we show how the degree of structure and randomness can be controlled to reduce statistical variance at the cost of increased computation and storage requirements. Empirical results strongly support our theory and justify the use of a broader family of structured matrices for scaling up kernel methods using random features.

We consider the problem of \textit{best arm identification} with a \textit{fixed budget $T$}, in the $K$-armed stochastic bandit setting, with arms distribution defined on $[0,1]$. We prove that any bandit strategy, for at least one bandit problem characterized by a complexity $H$, will misidentify the best arm with probability lower bounded by $$\exp\Big(-\frac{T}{\log(K)H}\Big),$$ where $H$ is the sum for all sub-optimal arms of the inverse of the squared gaps. Our result disproves formally the general belief - coming from results in the fixed confidence setting - that there must exist an algorithm for this problem whose probability of error is upper bounded by $\exp(-T/H)$. This also proves that some existing strategies based on the Successive Rejection of the arms are optimal - closing therefore the current gap between upper and lower bounds for the fixed budget best arm identification problem.

Our aim is to propose a model of subjective time based on information theory and to investigate its implications relative to two phenomena: 1 Time perception. Why does time appear to slow down when you visit a new place, and speed up once you get familiar with it? Recent findings in psychology, neuroscience, and ethology suggest that perceived duration does not coincide with physical duration, but rather depend on the statistical properties of stimuli. Experiments in psychophysics experiments have shown that, if presented with a train of repeated stimuli at constant time intervals (e.g., a letter, word, object, or face), subjects would perceive them as decreasing in duration [70, 44]. On the other hand, the opposite effect is reported whenever the properties of a train of stimuli are suddenly changed: brighter [10, 66], bigger [41, 73], dynamic [11, 31], or more complex stimuli [54, 49] appear to last longer. Measurements of brain activity have found that longer durations correlate with increased neuronal firing rates, fMRI, or EEG signals [2, 20, 16, 35, 45, 40]. When combined with ideas from information theory, these observations have led to the hypothesis that the subjective duration of a stimulus is proportional to the amount of neural energy required to represent said stimulus, and that this energy is a signature of the coding efficiency [21].