Kee Siong Ng

The MADlib analytics library: or MAD skills, the SQL

MADlib is a free, open-source library of in-database analytic methods. It provides an evolving suite of SQL-based algorithms for machine learning, data mining and statistics that run at scale within a database engine, with no need for data import/export to other tools. The goal is for MADlib to eventually serve a role for scalable database systems that is similar to the CRAN library for R: a community repository of statistical methods, this time written with scale and parallelism in mind. In this paper we introduce the MADlib project, including the background that led to its beginnings, and the motivation for its open-source nature. We provide an overview of the librarys architecture and design patterns, and provide a description of various statistical methods in that context. We include performance and speedup results of a core design pattern from one of those methods over the Greenplum parallel DBMS on a modest-sized test cluster. We then report on two initial efforts at incorporating academic research into MADlib, which is one of the projects goals. MADlib is freely available at http://madlib.net, and the project is open for contributions of both new methods, and ports to additional database platforms.

A Monte-Carlo AIXI approximation

This paper introduces a principled approach for the design of a scalable general reinforcement learning agent. Our approach is based on a direct approximation of AIXI, a Bayesian optimality notion for general reinforcement learning agents. Previously, it has been unclear whether the theory of AIXI could motivate the design of practical algorithms. We answer this hitherto open question in the affirmative, by providing the first computationally feasible approximation to the AIXI agent. To develop our approximation, we introduce a new Monte-Carlo Tree Search algorithm along with an agent-specific extension to the Context Tree Weighting algorithm. Empirically, we present a set of encouraging results on a variety of stochastic and partially observable domains. We conclude by proposing a number of directions for future research.

Personalisation for user agents

This paper is concerned with personalisation of user agents by symbolic, on-line machine learning techniques. The application of these ideas to an infotainment agent is discussed in detail. Also experimental results, which indicate that a high level of personalisation can be achieved by this approach, are presented.

Detecting Non-compliant Consumers in Spatio-Temporal Health Data: A Case Study from Medicare Australia

This paper describes our experience with applying data mining techniques to the problem of fraud detection in spatio-temporal health data in Medicare Australia. A modular framework that brings together disparate data mining techniques is adopted. Several generally applicable techniques for extracting features from spatial and temporal data are also discussed. The system was evaluated with input from domain experts and was found to achieve high hit rates. We also discuss some lessons drawn from the experience.

Probabilities on Sentences in an Expressive Logic

Abstract Automated reasoning about uncertain knowledge has many applications. One difficulty when developing such systems is the lack of a completely satisfactory integration of logic and probability. We address this problem directly. Expressive languages like higher-order logic are ideally suited for representing and reasoning about structured knowledge. Uncertain knowledge can be modeled by using graded probabilities rather than binary truth values. The main technical problem studied in this paper is the following: Given a set of sentences, each having some probability of being true, what probability should be ascribed to other (query) sentences? A natural wish-list, among others, is that the probability distribution (i) is consistent with the knowledge base, (ii) allows for a consistent inference procedure and in particular (iii) reduces to deductive logic in the limit of probabilities being 0 and 1, (iv) allows (Bayesian) inductive reasoning and (v) learning in the limit and in particular (vi) allows confirmation of universally quantified hypotheses/sentences. We translate this wish-list into technical requirements for a prior probability and show that probabilities satisfying all our criteria exist. We also give explicit constructions and several general characterizations of probabilities that satisfy some or all of the criteria and various (counter)examples. We also derive necessary and sufficient conditions for extending beliefs about finitely many sentences to suitable probabilities over all sentences, and in particular least dogmatic or least biased ones. We conclude with a brief outlook on how the developed theory might be used and approximated in autonomous reasoning agents. Our theory is a step towards a globally consistent and empirically satisfactory unification of probability and logic.

Probabilistic reasoning in a classical logic

We offer a view on how probability is related to logic. Specifically, we argue against the widely held belief that standard classical logics have no direct way of modelling the certainty of assumptions in theories and no direct way of stating the certainty of theorems proved from these (uncertain) assumptions. The argument rests on the observation that probability densities, being functions, can be represented and reasoned with naturally and directly in (classical) higher-order logic.

Learning Modal Theories

This paper discusses how to learn theories that are modal, concentrating on the issue of how modal hypotheses are formed. Illustrations are given to show the usefulness of the ideas for agent applications.

Probabilistic and Logical Beliefs

This paper proposes a method of integrating two different concepts of belief in artificial intelligence: belief as a probability distribution and belief as a logical formula. The setting for the integration is a highly expressive logic. The integration is explained in detail, as its comparison to other approaches to integrating logic and probability. An illustrative example is given to motivate the usefulness of the ideas in agent applications.

Declarative programming for agent applications

This paper introduces the execution model of a declarative programming language intended for agent applications. Features supported by the language include functional and logic programming idioms, higher-order functions, modal computation, probabilistic computation, and some theorem-proving capabilities. The need for these features is motivated and examples are given to illustrate the central ideas.

Reflections on agent beliefs

Some issues concerning beliefs of agents are discussed. These issues are the general syntactic form of beliefs, the logic underlying beliefs, acquiring beliefs, and reasoning with beliefs. The logical setting is more expressive and aspects of the reasoning and acquisition processes are more general than are usually considered.