In Chess, as in other popular strategic board games, players have different styles. For example, in Chess some players are more "positional" and other more "tactical", and this difference in style will affect their move choice in any given board position, and more generally their overall plan. The problem we tackle in this paper is that of applying machine learning to teach a computer to discriminate between players based on their style. Before we explain our methodology, we briefly review the method of temporal difference learning, which is central to our approach. Temporal difference learning [Sut88] is a machine learning technique, originating from the seminal work of Samuel [Sam59], in which learning occurs by minimising the differences between predictions and actual outcomes of a temporal sequence of observations. Samuel [Sam59] used the game of Checkers as a vehicle to study the feasibility of a computer learning from experience. Although the program written by Samuel did not achieve master strength, it was the precursor of the Checkers program Chinook [Sch97, SHJ01], which was the first computer program to win a match against a human world champion.

I argue that data becomes temporarily interesting by itself to some self-improving, but computationally limited, subjective observer once he learns to predict or compress the data in a better way, thus making it subjectively simpler and more beautiful. Curiosity is the desire to create or discover more non-random, non-arbitrary, regular data that is novel and surprising not in the traditional sense of Boltzmann and Shannon but in the sense that it allows for compression progress because its regularity was not yet known. This drive maximizes interestingness, the first derivative of subjective beauty or compressibility, that is, the steepness of the learning curve. It motivates exploring infants, pure mathematicians, composers, artists, dancers, comedians, yourself, and (since 1990) artificial systems.

Auctions are markets with strict regulations governing the information available to traders in the market and the possible actions they can take. Since well designed auctions achieve desirable economic outcomes, they have been widely used in solving real-world optimization problems, and in structuring stock or futures exchanges. Auctions also provide a very valuable testing-ground for economic theory, and they play an important role in computer-based control systems. Auction mechanism design aims to manipulate the rules of an auction in order to achieve specific goals. Economists traditionally use mathematical methods, mainly game theory, to analyze auctions and design new auction forms. However, due to the high complexity of auctions, the mathematical models are typically simplified to obtain results, and this makes it difficult to apply results derived from such models to market environments in the real world. As a result, researchers are turning to empirical approaches. This report aims to survey the theoretical and empirical approaches to designing auction mechanisms and trading strategies with more weights on empirical ones, and build the foundation for further research in the field.

Numerous studies have used coexpression of large expression datasets to infer functional associations between genes [1], to identify groups of related genes that are important in specific cancers or represent common tumour progression mechanisms [2], to study evolutionary change [3], for integration with other large-scale datasets [4][5], [6], and for the generation of high-quality biological interaction networks [7][8][9] [10]. A number of studies have also attempted to use coexpression to identify coregulation with the hypothesis that if two or more genes are expressed at the same time and location and at similar levels then they may be regulated by the same transcription factors and regulatory elements. This approach has shown promise particularly in simpler model organisms such as A. thaliana and S. cerevisiae [11] [12][13] [14] and many groups are currently working on implementing this idea in mammalian systems. However, traditional clustering methods have not worked particularly well on large datasets for this problem. Most methods assign each gene to only one cluster while in reality many genes likely take part in multiple processes. Also, global coexpression is measured across all conditions, whereas, it is probable that most genes are only tightly coregulated under certain conditions or locations. In recent years, a new field of clustering analysis termed subspace clustering (or biclustering) has gained increasing popularity in the analysis of gene expression data and other biological data [15][16][17][18] [19]. In contrast to traditional clustering methods such as hierarchical clustering, subspace clustering methods do not require expression to be correlated across all conditions for genes to be assigned to the same cluster. This has several advantages for data in which biologically relevant subsets exist (e.g.

This papers develops a logical language for representing probabilistic causal laws. Our interest in such a language is twofold. First, it can be motivated as a fundamental study of the representation of causal knowledge. Causality has an inherent dynamic aspect, which has been studied at the semantical level by Shafer in his framework of probability trees. In such a dynamic context, where the evolution of a domain over time is considered, the idea of a causal law as something which guides this evolution is quite natural. In our formalization, a set of probabilistic causal laws can be used to represent a class of probability trees in a concise, flexible and modular way. In this way, our work extends Shafer's by offering a convenient logical representation for his semantical objects. Second, this language also has relevance for the area of probabilistic logic programming. In particular, we prove that the formal semantics of a theory in our language can be equivalently defined as a probability distribution over the well-founded models of certain logic programs, rendering it formally quite similar to existing languages such as ICL or PRISM. Because we can motivate and explain our language in a completely self-contained way as a representation of probabilistic causal laws, this provides a new way of explaining the intuitions behind such probabilistic logic programs: we can say precisely which knowledge such a program expresses, in terms that are equally understandable by a non-logician. Moreover, we also obtain an additional piece of knowledge representation methodology for probabilistic logic programs, by showing how they can express probabilistic causal laws.

In this paper we apply computer learning methods to diagnosing ovarian cancer using the level of the standard biomarker CA125 in conjunction with information provided by mass-spectrometry. We are working with a new data set collected over a period of 7 years. Using the level of CA125 and mass-spectrometry peaks, our algorithm gives probability predictions for the disease. To estimate classification accuracy we convert probability predictions into strict predictions. Our algorithm makes fewer errors than almost any linear combination of the CA125 level and one peak's intensity (taken on the log scale). To check the power of our algorithm we use it to test the hypothesis that CA125 and the peaks do not contain useful information for the prediction of the disease at a particular time before the diagnosis. Our algorithm produces $p$-values that are better than those produced by the algorithm that has been previously applied to this data set. Our conclusion is that the proposed algorithm is more reliable for prediction on new data.

Many learning environments are able to store very detailed traces of students' activities thus producing huge sets of low-level data. However, identifying high-level behaviors from these data is not straightforward, especially if the concepts of the domain knowledge are not explicitly encoded together with the corresponding traces. In this paper we present a general approach that aims at discovering patterns of student behaviors. Its principles are applicable whenever the information carried by the traces may be split as finite sequences of {initial state, final state} pairs, where the final states are the result of basic student transformations performed on the corresponding initial states. Within this context, final states are the initial states of subsequent {initial state, final state} pairs (unless they are at the end of the sequence).

As ontologies proliferate and automatic reasoners become more powerful, the problem of protecting sensitive information becomes more serious. In particular, as facts can be inferred from other facts, it becomes increasingly likely that information included in an ontology, while not itself deemed sensitive, may be able to be used to infer other sensitive information. We first consider the problem of testing an ontology for safeness defined as its not being able to be used to derive any sensitive facts using a given collection of inference rules. We then consider the problem of optimizing an ontology based on the criterion of making as much useful information as possible available without revealing any sensitive facts.

Adequate representation of natural language semantics requires access to vast amounts of common sense and domain-specific world knowledge. Prior work in the field was based on purely statistical techniques that did not make use of background knowledge, on limited lexicographic knowledge bases such as WordNet, or on huge manual efforts such as the CYC project. Here we propose a novel method, called Explicit Semantic Analysis (ESA), for fine-grained semantic interpretation of unrestricted natural language texts. Our method represents meaning in a high-dimensional space of concepts derived from Wikipedia, the largest encyclopedia in existence. We explicitly represent the meaning of any text in terms of Wikipedia-based concepts. We evaluate the effectiveness of our method on text categorization and on computing the degree of semantic relatedness between fragments of natural language text. Using ESA results in significant improvements over the previous state of the art in both tasks. Importantly, due to the use of natural concepts, the ESA model is easy to explain to human users.

We study the regret of optimal strategies for online convex optimization games. Using von Neumann's minimax theorem, we show that the optimal regret in this adversarial setting is closely related to the behavior of the empirical minimization algorithm in a stochastic process setting: it is equal to the maximum, over joint distributions of the adversary's action sequence, of the difference between a sum of minimal expected losses and the minimal empirical loss. We show that the optimal regret has a natural geometric interpretation, since it can be viewed as the gap in Jensen's inequality for a concave functional--the minimizer over the player's actions of expected loss--defined on a set of probability distributions. We use this expression to obtain upper and lower bounds on the regret of an optimal strategy for a variety of online learning problems. Our method provides upper bounds without the need to construct a learning algorithm; the lower bounds provide explicit optimal strategies for the adversary.