Empirical evidence for ultrametric structure in multi-layer perceptron error surfaces

Combinatorial optimization problems share an interesting property with spin glass systems in that their state spaces can exhibit ultrametric structure. We use sampling methods to analyse the error surfaces of feedforward multi-layer perceptron neural networks learning encoder problems. The third order statistics of these points of attraction are examined and found to be arranged in a highly ultrametric way. This is a unique result for a finite, continuous parameter space. The implications of this result are discussed.

The philosophical significance of Cox's theorem

Cox's theorem states that, under certain assumptions, any measure of belief is isomorphic to a probability measure. This theorem, although intended as a justification of the subjectivist interpretation of probability theory, is sometimes presented as an argument for more controversial theses. Of particular interest is the thesis that the only coherent means of representing uncertainty is via the probability calculus. In this paper I examine the logical assumptions of Cox's theorem and I show how these impinge on the philosophical conclusions thought to be supported by the theorem. I show that the more controversial thesis is not supported by Cox's theorem. (C) 2003 Elsevier Inc. All rights reserved.

Two conceptions of reason

Two conceptions of reason are considered - the planning conception, embodied, for example, in rational choice theory and other familiar paradigms, and the improvisational conception, emerging from work on artificial intelligence and organization theory. Two illustrations are given of the problematic nature of the planning conception: ( 1) the inevitability of incompleteness in contracting and ( 2) the burdens of reason identified by John Rawls. Two diagnoses are provided for these infirmities: ( 1) the inexhaustibility of description and ( 2) the constructed nature of preferences and values. An alternative improvisational model is sketched and risk-spreading and bet-hedging are identified as two of its key technologies.

Face recognition robust to head pose from one sample image

Most face recognition systems only work well under quite constrained environments. In particular, the illumination conditions, facial expressions and head pose must be tightly controlled for good recognition performance. In 2004, we proposed a new face recognition algorithm, Adaptive Principal Component Analysis (APCA) [4], which performs well against both lighting variation and expression change. But like other eigenface-derived face recognition algorithms, APCA only performs well with frontal face images. The work presented in this paper is an extension of our previous work to also accommodate variations in head pose. Following the approach of Cootes et al, we develop a face model and a rotation model which can be used to interpret facial features and synthesize realistic frontal face images when given a single novel face image. We use a Viola-Jones based face detector to detect the face in real-time and thus solve the initialization problem for our Active Appearance Model search. Experiments show that our approach can achieve good recognition rates on face images across a wide range of head poses. Indeed recognition rates are improved by up to a factor of 5 compared to standard PCA.

Proceedings of the Eighth Australian and New Zealand Intelligent Information Systems Conference

These are the full proceedings of the conference.

Interaction between Normative Systems and Cognitive Agents in Temporal Modal Defeasible Logic

While some recent frameworks on cognitive agents addressed the combination of mental attitudes with deontic concepts, they commonly ignore the representation of time. An exception is [1]that manages also some temporal aspects both with respect to cognition and normative provisions. We propose in this paper an extension of the logic presented in [1]with temporal intervals.

Workgroup emotional intelligence: Scale development and relationship to team process effectiveness and goal focus

Over the last decade, ambitious claims have been made in the management literature about the contribution of emotional intelligence to success and performance. Writers in this genre have predicted that individuals with high emotional intelligence perform better in all aspects of management. This paper outlines the development of a new emotional intelligence measure, the Workgroup Emotional Intelligence Profile, Version 3 (WEIP-3), which was designed specifically to profile the emotional intelligence of individuals in work teams. We applied the scale in a study of the link between emotional intelligence and two measures of team performance: team process effectiveness and team goal focus. The results suggest that the average level of emotional intelligence of team members, as measured by the WEIP-3, is reflected in the initial performance of teams. In our study, low emotional intelligence teams initially performed at a lower level than the high emotional intelligence teams. Over time, however, teams with low average emotional intelligence raised their performance to match that of teams with high emotional intelligence.

Development and evaluation of neural network freeway incident detection models using field data

This paper discusses a multi-layer feedforward (MLF) neural network incident detection model that was developed and evaluated using field data. In contrast to published neural network incident detection models which relied on simulated or limited field data for model development and testing, the model described in this paper was trained and tested on a real-world data set of 100 incidents. The model uses speed, flow and occupancy data measured at dual stations, averaged across all lanes and only from time interval t. The off-line performance of the model is reported under both incident and non-incident conditions. The incident detection performance of the model is reported based on a validation-test data set of 40 incidents that were independent of the 60 incidents used for training. The false alarm rates of the model are evaluated based on non-incident data that were collected from a freeway section which was video-taped for a period of 33 days. A comparative evaluation between the neural network model and the incident detection model in operation on Melbourne's freeways is also presented. The results of the comparative performance evaluation clearly demonstrate the substantial improvement in incident detection performance obtained by the neural network model. The paper also presents additional results that demonstrate how improvements in model performance can be achieved using variable decision thresholds. Finally, the model's fault-tolerance under conditions of corrupt or missing data is investigated and the impact of loop detector failure/malfunction on the performance of the trained model is evaluated and discussed. The results presented in this paper provide a comprehensive evaluation of the developed model and confirm that neural network models can provide fast and reliable incident detection on freeways. (C) 1997 Elsevier Science Ltd. All rights reserved.