We present an unsupervised technique for visual learning which is based on density estimation in high-dimensional spaces using an eigenspace decomposition. Two types of density estimates are derived for modeling the training data: a multivariate Gaussian (for unimodal distributions) and a Mixture-of-Gaussians model (for multimodal distributions). These probability densities are then used to formulate a maximum-likelihood estimation framework for visual search and target detection for automatic object recognition and coding. Our learning technique is applied to the probabilistic visual modeling, detection, recognition, and coding of human faces and non-rigid objects such as hands.

Images containing faces are essential to intelligent vision-based human computer interaction, and research efforts in face processing include face recognition, face tracking, pose estimation, and expression recognition. However, many reported methods assume that the faces in an image or an image sequence have been identified and localized. To build fully automated systems that analyze the information contained in face images, robust and efficient face detection algorithms are required. Given a single image, the goal of face detection is to identify all image regions which contain a face regardless of its three-dimensional position, orientation, and the lighting conditions. Such a problem is challenging because faces are nonrigid and have a high degree of variability in size, shape, color, and texture. Numerous techniques have been developed to detect faces in a single image, and the purpose of this paper is to categorize and evaluate these algorithms. We also discuss relevant issues such as data collection, evaluation metrics, and benchmarking. After analyzing these algorithms and identifying their limitations, we conclude with several promising directions for future research.

We present an unsupervised technique for visual learning which is based on density estimation in high-dimensional spaces using an eigenspace decomposition. Two types of density estimates are derived for modeling the training data: a multivariate Gaussian (for a unimodal distribution) and a multivariate Mixture-of-Gaussians model (for multimodal distributions). These probability densities are then used to formulate a maximum-likelihood estimation framework for visual search and target detection for automatic object recognition. This learning technique is tested in experiments with modeling and subsequent detection of human faces and non-rigid objects such as hands.

This note derives the stationary behavior of idealized TCP congestion avoidance. More specifically, it derives the stationary distribution of the congestion window size if loss of packets are independentevents with equal probability. The mathematical derivation uses a fluid flow, continuous time, approximation to the discrete time process #W n #, where W n is the congestion window after the n-th packet. We derive explicit results for the stationary distribution and all its moments. Congestion avoidance is the algorithm used by TCP to set its window size (and indirectly its data rate) under moderate to light segment (packet) losses. The congestion avoidance mechanism we model is idealized in the sense that loss of multiple packets does not lead to time-out phenomena. Such idealized behavior can be implemented using Selective Acknowledgements (SACKs). As such, our model predicts behavior of TCP with SACKs. It also is an approximate model in other situations. Among the results are that if eve...

In this paper we propose two methods for separating mixtures of independent sources without any precise knowledge of their probability distribution. They are obtained by considering a maximum likelihood solution corresponding to some given distributions of the sources and relaxing this assumption afterward. The first method is specially adapted to temporally independent non Gaussian sources and is based on the use of nonlinear separating functions. The second method is specially adapted to correlated sources with distinct spectra and is based on the use of linear separating filters. A theoretical analysis of the performance of the methods has been made. A simple procedure for choosing optimally the separating functions from a given linear space of functions is proposed. Further, in the second method, a simple implementation based on the simultaneous diagonalization of two symmetric matrices is provided. Finally, some numerical and simulation results are given illustrating the performan...

Abstract — In this paper, we investigate a state estimation problem involving finite communication capacity constraints. Unlike classical estimation problems where the observation is a continuous process corrupted by additive noises, there is a constraint that the observations must be coded and transmitted over a digital communication channel with finite capacity. This problem is formulated mathematically, and some convergence properties are defined. Moreover, the concept of a finitely recursive coder-estimator sequence is introduced. A new upper bound for the average estimation error is derived for a large class of random variables. Convergence properties of some coder-estimator algorithms are analyzed. Various conditions connecting the communication data rate with the rate of change of the underlying dynamics are established for the existence of stable and asymptotically convergent coder-estimator schemes. Index Terms—Finitely recursive coder-estimator sequence, hybrid systems, prefix code, state estimation. I.

The human visual system is proficient in perceiving three-dimensional shape from the shading patterns in a two-dimensional image. How it does this is not well understood and continues to be a question of fundamental and practical interest. In this paper we present a new quantitative approach to shape-from-shading that may provide some answers. We suggest that the brain, through evolution or prior experience, has discovered that objects can be classified into lower-dimensional object-classes as to their shape. Extraction of shape from shading is then equivalent to the much simpler problem of parameter estimation in a low dimensional space. We carry out this proposal for an important class of 3D objects; human heads. From an ensemble of several hundred laser-scanned 3D heads, we use principal component analysis to derive a low-dimensional parameterization of head shape space. An algorithm for solving shape-from-shading using this representation is presented. It works well even on real im...

We consider packet routing when packets are injected continuously into a network. We develop an adversarial theory of queuing aimed at addressing some of the restrictions inherent in probabilistic analysis and queuing theory based on time-invariant stochastic generation. We examine the stability of queuing networks and policies when the arrival process is adversarial, and provide some preliminary results in this direction. Our approach sheds light on various queuing policies in simple networks, and paves the way for a systematic study of queuing with few or no probabilistic assumptions.

The need for unified notation in atmospheric and oceanic data assimilation arises from the field's rapid theoretical expansion and the desire to translate it into practical applications. Self-consistent notation is proposed that bridges sequential and variational methods, on the one hand, and operational usage, on the other. Over various other mottoes for this risky endeavor, the authors selected: "When I use a word," Humpty Dumpty said, in rather a scornful voice tone, "it means just what I choose it to mean --- neither more nor less." Lewis Carroll, 1871. 1 J. Met. Soc. Japan, Special Issue on "Data Assimilation in Meteorology and Oceanography: Theory and Practice." Vol. 75, No. 1B, pp. 181--189, 1997. 2 Corresponding author. 3 Current affiliation: CNES, 2, place Maurice Quentin, 75039 Paris Cedex 01, France. 1 J. Met. Soc. Japan, (1997), K. Ide, P. Courtier, M. Ghil and A.C. Lorenc 2 1. Introduction and motivation Model-based assimilation of observations, or data assimilati...

Current solutions to providing statistical performance guarantees to bursty traffic such as compressed video encounter several problems: 1) source traffic descriptors are often too simple to capture the burstiness and important time-correlations of VBR sources or too complex to be used for admission control algorithms; 2) stochastic descriptions of a source are inherently difficult for the network to enforce or police; 3) multiplexing inside the network's queues may change the stochastic properties of the source in an intractable way, precluding the provision of end-toend QoS guarantees to heterogeneous sources with different performance requirements. In this paper, we present a new approach to providing end-to-end statistical performance guarantees that overcomes these limitations. We term the approach Hybrid Bounding Interval Dependent (H-BIND) because it uses the Deterministic-BIND traffic model to capture the correlation structure and burstiness properties of a stream; but unlike a...