Computer Science Division, University of California, Berkeley, CA 94720 Dept. of Physics & MIT Kavli Institute, Massachusetts Institute of Technology, Cambridge, MA 02139 and Future of Humanity Institute, Oxford University, 16-17 St. Ebbe's str., Oxford OX1 1PT, UK Success in the quest for artificial intelligence has the potential to bring unprecedented benefits to humanity, and it is therefore worthwhile to investigate how to maximize these benefits while avoiding potential pitfalls. This article gives numerous examples (which should by no means be construed as an exhaustive list) of such worthwhile research aimed at ensuring that AI remains robust and beneficial. Artificial intelligence (AI) research has explored a variety of problems and approaches since its inception, but for the last 20 years or so has been focused on the problems surrounding the construction of intelligent agents - systems that perceive and act in some environment. In this context, the criterion for intelligence is related to statistical and economic notions of rationality-colloquially, the ability to make good decisions, plans, or inferences. The adoption of probabilistic representations and statistical learning methods has led to a large degree of integration and cross-fertilization between AI, machine learning, statistics, control theory, neuroscience, and other fields. The establishment of shared theoretical frameworks, combined with the availability of data and processing power, has yielded remarkable successes in various component tasks such as speech recognition, image classification, autonomous vehicles, machine translation, legged locomotion, and question-answering systems. As capabilities in these areas and others cross the threshold from laboratory research to economically valuable technologies, a virtuous cycle takes hold whereby even small improvements in performance are worth large sums of money, prompting greater investments in research. There is now a broad consensus that AI research is progressing steadily, and that its impact on society is likely to increase.

Kernelized Support Vector Machines (SVMs) are among the best performing supervised learning methods. But for optimal predictive performance, time-consuming parameter tuning is crucial, which impedes application. To tackle this problem, the classic model selection procedure based on grid-search and cross-validation was refined, e.g. by data subsampling and direct search heuristics. Here we focus on a different aspect, the stopping criterion for SVM training. We show that by limiting the training time given to the SVM solver during parameter tuning we can reduce model selection times by an order of magnitude.

Supervised, semi-supervised, and unsupervised learning estimate a function given input/output samples. Generalization of the learned function to unseen data can be improved by incorporating side information into learning. Side information are data that are neither from the input space nor from the output space of the function, but include useful information for learning it. In this paper we show that learning with side information subsumes a variety of related approaches, e.g. multi-task learning, multi-view learning and learning using privileged information. Our main contributions are (i) a new perspective that connects these previously isolated approaches, (ii) insights about how these methods incorporate different types of prior knowledge, and hence implement different patterns, (iii) facilitating the application of these methods in novel tasks, as well as (iv) a systematic experimental evaluation of these patterns in two supervised learning tasks.

Peer grading is the process of students reviewing each others' work, such as homework submissions, and has lately become a popular mechanism used in massive open online courses (MOOCs). Intrigued by this idea, we used it in a course on algorithms and data structures at the University of Hamburg. Throughout the whole semester, students repeatedly handed in submissions to exercises, which were then evaluated both by teaching assistants and by a peer grading mechanism, yielding a large dataset of teacher and peer grades. We applied different statistical and machine learning methods to aggregate the peer grades in order to come up with accurate final grades for the submissions (supervised and unsupervised, methods based on numeric scores and ordinal rankings). Surprisingly, none of them improves over the baseline of using the mean peer grade as the final grade. We discuss a number of possible explanations for these results and present a thorough analysis of the generated dataset.

The random cluster model is used to define an upper bound on a distance measure as a function of the number of data points to be classified and the expected value of the number of classes to form in a hybrid K-means and regression classification methodology, with the intent of detecting anomalies. Conditions are given for the identification of classes which contain anomalies and individual anomalies within identified classes. A neural network model describes the decision region-separating surface for offline storage and recall in any new anomaly detection.

We study how to obtain concise descriptions of discrete multivariate sequential data. In particular, how to do so in terms of rich multivariate sequential patterns that can capture potentially highly interesting (cor)relations between sequences. To this end we allow our pattern language to span over the domains (alphabets) of all sequences, allow patterns to overlap temporally, as well as allow for gaps in their occurrences. We formalise our goal by the Minimum Description Length principle, by which our objective is to discover the set of patterns that provides the most succinct description of the data. To discover high-quality pattern sets directly from data, we introduce DITTO, a highly efficient algorithm that approximates the ideal result very well. Experiments show that DITTO correctly discovers the patterns planted in synthetic data. Moreover, it scales favourably with the length of the data, the number of attributes, the alphabet sizes. On real data, ranging from sensor networks to annotated text, DITTO discovers easily interpretable summaries that provide clear insight in both the univariate and multivariate structure.

Our world is filled with both beautiful and brainy people, but how often does a Nobel Prize winner also wins a beauty pageant? Let us assume that someone who is both very beautiful and very smart is more rare than what we would expect from the combination of the number of beautiful and brainy people. Of course there will still always be some individuals that defy this stereotype; these beautiful brainy people are exactly the class of anomaly we focus on in this paper. They do not posses intrinsically rare qualities, it is the unexpected combination of factors that makes them stand out. In this paper we define the above described class of anomaly and propose a method to quickly identify them in transaction data. Further, as we take a pattern set based approach, our method readily explains why a transaction is anomalous. The effectiveness of our method is thoroughly verified with a wide range of experiments on both real world and synthetic data.

Gesture recognition is one of the core components in the thriving research field of humancomputer interaction. The recognition of distinct hand and arm motions is becoming increasingly important, as it enables smart interactions with electronic devices. Furthermore, gesture identification in video can be seen as a first step towards sign language recognition, where even subtle differences in motion can play an important role. Some examples that complicate the identification of gestures are changes in background and lighting due to the varying environment, variations in the performance and speed of the gestures, different clothes worn by the performers and different positioning relative to the camera. Moreover, regular hand motion or out-of-vocabulary gestures should not to be confused with one of the target gestures. Convolutional neural networks (CNNs) (LeCun et al., 1998) are the de facto standard approach in computer vision. CNNs have the ability to learn complex hierarchies with increasing levels of abstraction while being end-to-end trainable. Their success has had a huge impact on vision based applications like image classification (Krizhevsky et al., 2012), object detection

Given a Markov Decision Process (MDP) with $n$ states and a totalnumber $m$ of actions, we study the number of iterations needed byPolicy Iteration (PI) algorithms to converge to the optimal$\gamma$-discounted policy. We consider two variations of PI: Howard'sPI that changes the actions in all states with a positive advantage,and Simplex-PI that only changes the action in the state with maximaladvantage. We show that Howard's PI terminates after at most $O\left(\frac{m}{1-\gamma}\log\left(\frac{1}{1-\gamma}\right)\right)$iterations, improving by a factor $O(\log n)$ a result by Hansen etal., while Simplex-PI terminates after at most $O\left(\frac{nm}{1-\gamma}\log\left(\frac{1}{1-\gamma}\right)\right)$iterations, improving by a factor $O(\log n)$ a result by Ye. Undersome structural properties of the MDP, we then consider bounds thatare independent of the discount factor~$\gamma$: quantities ofinterest are bounds $\tau\_t$ and $\tau\_r$---uniform on all states andpolicies---respectively on the \emph{expected time spent in transientstates} and \emph{the inverse of the frequency of visits in recurrentstates} given that the process starts from the uniform distribution.Indeed, we show that Simplex-PI terminates after at most $\tilde O\left(n^3 m^2 \tau\_t \tau\_r \right)$ iterations. This extends arecent result for deterministic MDPs by Post & Ye, in which $\tau\_t\le 1$ and $\tau\_r \le n$, in particular it shows that Simplex-PI isstrongly polynomial for a much larger class of MDPs. We explain whysimilar results seem hard to derive for Howard's PI. Finally, underthe additional (restrictive) assumption that the state space ispartitioned in two sets, respectively states that are transient andrecurrent for all policies, we show that both Howard's PI andSimplex-PI terminate after at most $\tilde O(m(n^2\tau\_t+n\tau\_r))$iterations.

We consider the problem of configuration formation in modular robot systems where a set of modules that are initially in different configurations and located at different locations are required to assume appropriate positions so that they can get into a new, user-specified, target configuration. We propose a novel algorithm based on graph isomorphism, where the modules select locations or spots in the target configuration using a utility-based framework, while retaining their original configuration to the greatest extent possible, to reduce the time and energy required by the modules to assume the target configuration. We have shown analytically that our proposed algorithm is complete and guarantees a Pareto-optimal allocation. Experimental simulations of our algorithm with different number of modules in different initial configurations and located initially at different locations, show that the planning time of our algorithm is nominal (order of msec. for 100 modules). We have also compared our algorithm against a market-based allocation algorithm and shown that our proposed algorithm performs better in terms of time and number of messages exchanged.