We present a graphical model for beat tracking in recorded music. Using a probabilistic graphical model allows us to incorporate local information and global smoothness constraints in a principled manner. We evaluate our model on a set of varied and difficult examples, and achieve impressive results. By using a fast dual-tree algorithm for graphical model inference, our system runs in less time than the duration of the music being processed.

We present a graphical model for beat tracking in recorded music. Using a probabilistic graphical model allows us to incorporate local information and global smoothness constraints in a principled manner. We evaluate our model on a set of varied and difficult examples, and achieve impressive results. By using a fast dual-tree algorithm for graphical model inference, our system runs in less time than the duration of the music being processed.

We describe how we used a data set of chorale harmonisations composed by Johann Sebastian Bach to train Hidden Markov Models. Using a probabilistic framework allows us to create a harmonisation system which learns from examples, and which can compose new harmonisations. We make a quantitative comparison of our system's harmonisation performance against simpler models, and provide example harmonisations.

Section 2 below gives an overview of the musical background to chorale harmonisation. Section 3 explains how we can create a harmonisation system using Hidden Markov Models. Section 4 examines the system's performance quantitatively and provides example

Dixon describes beats as follows: "much music has as its rhythmic basis a series of pulses, spaced approximately equally in time, relative to which the timing of all musical events can be described. This phenomenon is called the beat, and the individual pulses are also called beats"[1]. Given a piece of recorded music (an MP3 file, for example), we wish to produce a set of beats that correspond to the beats perceived by human listeners. The set of beats of a song can be characterised by the trajectories through time of thetempo and phase offset. Tempo is typically measured in beats per minute (BPM), and describes the frequency of beats.

The context in which a name appears in a caption provides powerful cues as to who is depicted in the associated image. We obtain 44,773 face images, usinga face detector, from approximately half a million captioned news images and automatically link names, obtained using a named entity recognizer,with these faces. A simple clustering method can produce fairresults. We improve these results significantly by combining the clustering process with a model of the probability that an individual is depicted given its context. Once the labeling procedure is over, we have an accurately labeled set of faces, an appearance model for each individual depicted, and a natural language model that can produce accurate resultson captions in isolation.

The Graphplan algorithm for generating optimal make-span plans containing parallel sets of actions remains one of the most effective ways to generate such plans. However, despite enhancements on a range of fronts, the approach is currently dominated in terms of speed, by state space planners that employ distance-based heuristics to quickly generate serial plans. We report on a family of strategies that employ available memory to construct a search trace so as to learn from various aspects of Graphplan's iterative search episodes in order to expedite search in subsequent episodes. The planning approaches can be partitioned into two classes according to the type and extent of search experience captured in the trace. The planners using the more aggressive tracing method are able to avoid much of Graphplan's redundant search effort, while planners in the second class trade off this aspect in favor of a much higher degree of freedom than Graphplan in traversing the space of'states' generated during regression search on the planning graph. The tactic favored by the second approach, exploiting the search trace to transform the depth-first, IDA* nature of Graphplan's search into an iterative state space view, is shown to be the more powerful. We demonstrate that distance-based, state space heuristics can be adapted to informed traversal of the search trace used by the second class of planners and develop an augmentation targeted specifically at planning graph search. Guided by such a heuristic, the step-optimal version of the planner in this class clearly dominates even a highly enhanced version of Graphplan. By adopting beam search on the search trace we then show that virtually optimal parallel plans can be generated at speeds quite competitive with a modern heuristic state space planner.

Blur is one of the most common forms of image distortion. It can arise from a variety of sources, such as atmospheric scatter, lens defocus, optical aberrations of the lens, and spatial and temporal sensor integration. Human observers are bothered by blur, and our visual systems are quite good at reporting whether an image appears blurred (or sharpened) [1, 2]. However, the mechanism by which this is accomplished is not well understood. Clearly, detection of blur requires some model of what constitutes an unblurred image. In recent years, there has been a surge of interest in the modelling of natural images, both for purposes of improving the performance of image processing and computer vision systems, and also for furthering our understanding of biological visual systems.

The problem of "Structure From Motion" is a central problem in vision: given the 2D locations of certain points we wish to recover the camera motion and the 3D coordinates of the points. Under simplified camera models, the problem reduces to factorizing a measurement matrix into the product of two low rank matrices. Each element of the measurement matrix contains the position of a point in a particular image. When all elements are observed, the problem can be solved trivially using SVD, but in any realistic situation many elements of the matrix are missing and the ones that are observed have a different directional uncertainty. Under these conditions, most existing factorization algorithms fail while human perception is relatively unchanged. In this paper we use the well known EM algorithm for factor analysis to perform factorization. This allows us to easily handle missing data and measurement uncertainty and more importantly allows us to place a prior on the temporal trajectory of the latent variables (the camera position). We show that incorporating this prior gives a significant improvement in performance in challenging image sequences.

This paper applies fast sparse multidimensional scaling (MDS) to a large graph of music similarity, with 267K vertices that represent artists, albums, and tracks; and 3.22M edges that represent similarity between those entities. Once vertices are assigned locations in a Euclidean space, the locations can be used to browse music and to generate playlists. MDS on very large sparse graphs can be effectively performed by a family of algorithms called Rectangular Dijsktra (RD) MDS algorithms. These RD algorithms operate on a dense rectangular slice of the distance matrix, created by calling Dijsktra a constant number of times. Two RD algorithms are compared: Landmark MDS, which uses the Nyström approximation to perform MDS; and a new algorithm called Fast Sparse Embedding, which uses FastMap. These algorithms compare favorably to Laplacian Eigenmaps, both in terms of speed and embedding quality.