We compare discriminative and generative learning as typified by logistic regression and naive Bayes. We show, contrary to a widely(cid:173) held belief that discriminative classifiers are almost always to be preferred, that there can often be two distinct regimes of per(cid:173) formance as the training set size is increased, one in which each algorithm does better. This stems from the observation- which is borne out in repeated experiments- that while discriminative learning has lower asymptotic error, a generative classifier may also approach its (higher) asymptotic error much faster.

Acetylcholine (ACh) has been implicated in a wide variety of tasks involving attentional processes and plasticity. Following extensive animal studies, it has previously been suggested that ACh reports on uncertainty and controls hippocampal, cortical and cortico-amygdalar plasticity. We extend this view and consider its effects on cortical representational inference, arguing that ACh controls the balance between bottom-up inference, in(cid:3)uenced by input stimuli, and top-down inference, in(cid:3)uenced by contextual information. We illustrate our proposal using a hierarchical hid- den Markov model.

We present a principled and efficient planning algorithm for cooperative multia- gent dynamic systems. A striking feature of our method is that the coordination and communication between the agents is not imposed, but derived directly from the system dynamics and function approximation architecture. We view the en- tire multiagent system as a single, large Markov decision process (MDP), which we assume can be represented in a factored way using a dynamic Bayesian net- work (DBN). The action space of the resulting MDP is the joint action space of the entire set of agents. Our approach is based on the use of factored linear value functions as an approximation to the joint value function.

We derive an equivalence between AdaBoost and the dual of a convex optimization problem, showing that the only difference between mini- mizing the exponential loss used by AdaBoost and maximum likelihood for exponential models is that the latter requires the model to be normal- ized to form a conditional probability distribution over labels. In addi- tion to establishing a simple and easily understood connection between the two methods, this framework enables us to derive new regularization procedures for boosting that directly correspond to penalized maximum likelihood. Experiments on UCI datasets support our theoretical analy- sis and give additional insight into the relationship between boosting and logistic regression.

In reinforcement learning problems, the aim is to select a controller that will maximize the average reward in some environment.

We describe a computer system that provides a real-time musi(cid:173) cal accompaniment for a live soloist in a piece of non-improvised music for soloist and accompaniment. A Bayesian network is devel(cid:173) oped that represents the joint distribution on the times at which the solo and accompaniment notes are played, relating the two parts through a layer of hidden variables. The network is first con(cid:173) structed using the rhythmic information contained in the musical score. The network is then trained to capture the musical interpre(cid:173) tations of the soloist and accompanist in an off-line rehearsal phase. During live accompaniment the learned distribution of the network is combined with a real-time analysis of the soloist's acoustic sig(cid:173) nal, performed with a hidden Markov model, to generate a musi(cid:173) cally principled accompaniment that respects all available sources of knowledge.

We show that states of a dynamical system can be usefully repre(cid:173) sented by multi-step, action-conditional predictions of future ob(cid:173) servations. State representations that are grounded in data in this way may be easier to learn, generalize better, and be less depen(cid:173) dent on accurate prior models than, for example, POMDP state representations. Building on prior work by Jaeger and by Rivest and Schapire, in this paper we compare and contrast a linear spe(cid:173) cialization of the predictive approach with the state representa(cid:173) tions used in POMDPs and in k-order Markov models. Ours is the first specific formulation of the predictive idea that includes both stochasticity and actions (controls). We show that any system has a linear predictive state representation with number of predictions no greater than the number of states in its minimal POMDP model.

This paper proposes an approach to classification of adjacent segments of a time series as being either of classes. We use a hierarchical model that consists of a feature extraction stage and a generative classifier which is built on top of these features. Such two stage approaches are often used in signal and image processing. The novel part of our work is that we link these stages probabilistically by using a latent feature space. To use one joint model is a Bayesian requirement, which has the advantage to fuse information according to its certainty.

Statistical learning and probabilistic inference techniques are used to in- fer the hand position of a subject from multi-electrode recordings of neu- ral activity in motor cortex. First, an array of electrodes provides train- ing data of neural firing conditioned on hand kinematics. We learn a non- parametric representation of this firing activity using a Bayesian model and rigorously compare it with previous models using cross-validation. Second, we infer a posterior probability distribution over hand motion conditioned on a sequence of neural test data using Bayesian inference. The learned firing models of multiple cells are used to define a non- Gaussian likelihood term which is combined with a prior probability for the kinematics.

The recent introduction of the'relevance vector machine' has effec(cid:173) tively demonstrated how sparsity may be obtained in generalised linear models within a Bayesian framework. Using a particular form of Gaussian parameter prior, 'learning' is the maximisation, with respect to hyperparameters, of the marginal likelihood of the data. This paper studies the properties of that objective func(cid:173) tion, and demonstrates that conditioned on an individual hyper(cid:173) parameter, the marginal likelihood has a unique maximum which is computable in closed form. It is further shown that if a derived'sparsity criterion' is satisfied, this maximum is exactly equivalent to'pruning' the corresponding parameter from the model.