We report on the staging of the first competition on knowledge engineering for AI planning and scheduling systems, held in Monterey, California, in colocation with the ICAPS 2005 conference. The background and motivation is discussed, together with the relationship of this new competition with the current international planning competition. We report on the new competition's format, its outcome, and the benefits we hope it will bring to the research area.

In a peer-to-peer inference system, each peer can reason locally but can also solicit some of its acquaintances, which are peers sharing part of its vocabulary. In this paper, we consider peer-to-peer inference systems in which the local theory of each peer is a set of propositional clauses defined upon a local vocabulary. An important characteristic of peer-to-peer inference systems is that the global theory (the union of all peer theories) is not known (as opposed to partition-based reasoning systems). The main contribution of this paper is to provide the first consequence finding algorithm in a peer-to-peer setting: DeCA. It is anytime and computes consequences gradually from the solicited peer to peers that are more and more distant. We exhibit a sufficient condition on the acquaintance graph of the peer-to-peer inference system for guaranteeing the completeness of this algorithm. Another important contribution is to apply this general distributed reasoning setting to the setting of the Semantic Web through the Somewhere semantic peer-to-peer data management system. The last contribution of this paper is to provide an experimental analysis of the scalability of the peer-to-peer infrastructure that we propose, on large networks of 1000 peers.

Device mismatch in VLSI degrades the accuracy of analog arithmetic circuits and lowers the learning performance of large-scale neural networks implemented in this technology. We show compact, low-power on-chip calibration techniques that compensate for device mismatch. Our techniques enable large-scale analog VLSI neural networks with learning performance on the order of 10 bits. We demonstrate our techniques on a 64-synapse linear perceptron learning with the Least-Mean-Squares (LMS) algorithm, and fabricated in a 0.35µm CMOS process.

Device mismatch in VLSI degrades the accuracy of analog arithmetic circuits and lowers the learning performance of large-scale neural networks implemented in this technology. We show compact, low-power on-chip calibration techniques that compensate for device mismatch. Our techniques enable large-scale analog VLSI neural networks with learning performance on the order of 10 bits. We demonstrate our techniques on a 64-synapse linear perceptron learning with the Least-Mean-Squares (LMS) algorithm, and fabricated in a 0.35µm CMOS process.

Device mismatch in VLSI degrades the accuracy of analog arithmetic circuits and lowers the learning performance of large-scale neural networks implementedin this technology. We show compact, low-power on-chip calibration techniques that compensate for device mismatch. Our techniques enable large-scale analog VLSI neural networks with learning performanceon the order of 10 bits. We demonstrate our techniques on a 64-synapse linear perceptron learning with the Least-Mean-Squares (LMS) algorithm, and fabricated in a 0.35µm CMOS process.

We describe the version of the GPT planner used in the probabilistic track of the 4th International Planning Competition (ipc-4). This version, called mGPT, solves Markov Decision Processes specified in the ppddl language by extracting and using different classes of lower bounds along with various heuristic-search algorithms. The lower bounds are extracted from deterministic relaxations where the alternative probabilistic effects of an action are mapped into different, independent, deterministic actions. The heuristic-search algorithms use these lower bounds for focusing the updates and delivering a consistent value function over all states reachable from the initial state and the greedy policy.

In this paper, we confront the problem of applying reinforcement learning to agents that perceive the environment through many sensors and that can perform parallel actions using many actuators as is the case in complex autonomous robots. We argue that reinforcement learning can only be successfully applied to this case if strong assumptions are made on the characteristics of the environment in which the learning is performed, so that the relevant sensor readings and motor commands can be readily identified. The introduction of such assumptions leads to strongly-biased learning systems that can eventually lose the generality of traditional reinforcement-learning algorithms. In this line, we observe that, in realistic situations, the reward received by the robot depends only on a reduced subset of all the executed actions and that only a reduced subset of the sensor inputs (possibly different in each situation and for each action) are relevant to predict the reward. We formalize this property in the so called 'categorizability assumption' and we present an algorithm that takes advantage of the categorizability of the environment, allowing a decrease in the learning time with respect to existing reinforcement-learning algorithms. Results of the application of the algorithm to a couple of simulated realistic-robotic problems (landmark-based navigation and the six-legged robot gait generation) are reported to validate our approach and to compare it to existing flat and generalization-based reinforcement-learning approaches.

Typically, translation algorithms from a symbolic to a connectionist representation and vice-versa are employed to provide either (i) a neural implementation of a logic, (ii) a logical characterisation of a neural system, or (iii) a hybrid learning system that brings together features from connectionism and symbolic artificial intelligence (Holldobler, 1993). Until recently, neural-symbolic systems were not able to fully represent, reason and learn expressive languages other than propositional and fragments of first-order logic (Cloete & Zurada, 2000). However, in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), a new approach to knowledge representation and reasoning in neural-symbolic systems based on neural networks ensembles has been introduced. This new approach shows that modal logics can be effectively represented in artificial neural networks. In this paper, following the approach introduced in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), we move one step further and show that temporal logics can be effectively represented in artificial neural o Artur Garcez is partly supported by the Nuffield Foundation. Luis Lamb is partly supported by CNPq. The authors would like to thank the referees for their comments.

Typically, translation algorithms from a symbolic to a connectionist representation and vice-versa are employed to provide either (i) a neural implementation of a logic, (ii) a logical characterisation of a neural system, or (iii) a hybrid learning system that brings together features from connectionism and symbolic artificial intelligence (Holldobler, 1993). Until recently, neural-symbolic systems were not able to fully represent, reason and learn expressive languages other than propositional and fragments of first-order logic (Cloete & Zurada, 2000). However, in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), a new approach to knowledge representation and reasoning in neural-symbolic systems based on neural networks ensembles has been introduced. This new approach shows that modal logics can be effectively represented in artificial neural networks. In this paper, following the approach introduced in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), we move one step further and show that temporal logics can be effectively represented in artificial neural o Artur Garcez is partly supported by the Nuffield Foundation. Luis Lamb is partly supported by CNPq. The authors would like to thank the referees for their comments.

This article reports on the RoboCup-2003 event. RoboCup is no longer just the Soccer World Cup for autonomous robots but has evolved to become a coordinated initiative encompassing four different robotics events: (1) Soccer, (2) Rescue, (3) Junior (focused on education), and (4) a Scientific Symposium. RoboCup-2003 took place from 2 to 11 July 2003 in Padua (Italy); it was colocated with other scientific events in the field of AI and robotics. In this article, in addition to reporting on the results of the games, we highlight the robotics and AI technologies exploited by the teams in the different leagues and describe the most meaningful scientific contributions.