Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Fundamentals Of Statistical Signal Processing

Monte Carlo methods are revolutionizing the on-line analysis of data in fields as diverse as financial modeling, target tracking and computer vision. These methods, appearing under the names of bootstrap filters, condensation, optimal Monte Carlo filters, particle filters and survival of the fittest, have made it possible to solve numerically many complex, non-standard problems that were previously intractable. This book presents the first comprehensive treatment of these techniques, including convergence results and applications to tracking, guidance, automated target recognition, aircraft navigation, robot navigation, econometrics, financial modeling, neural networks, optimal control, optimal filtering, communications, reinforcement learning, signal enhancement, model averaging and selection, computer vision, semiconductor design, population biology, dynamic Bayesian networks, and time series analysis. This will be of great value to students, researchers and practitioners, who have some basic knowledge of probability. Arnaud Doucet received the Ph. D. degree from the University of Paris-XI Orsay in 1997. From 1998 to 2000, he conducted research at the Signal Processing Group of Cambridge University, UK. He is currently an assistant professor at the Department of Electrical Engineering of Melbourne University, Australia. His research interests include Bayesian statistics, dynamic models and Monte Carlo methods. Nando de Freitas obtained a Ph.D. degree in information engineering from Cambridge University in 1999. He is presently a research associate with the artificial intelligence group of the University of California at Berkeley. His main research interests are in Bayesian statistics and the application of on-line and batch Monte Carlo methods to machine learning. Neil Gordon obtained a Ph.D. in Statistics from Imperial College, University of London in 1993. He is with the Pattern and Information Processing group at the Defence Evaluation and Research Agency in the United Kingdom. His research interests are in time series, statistical data analysis, and pattern recognition with a particular emphasis on target tracking and missile guidance.

/pdf/sequential-monte-carlo-methods-in-practice-3kfxeebn8e.pdf

Sequential Monte Carlo methods in practice

物件導向軟體之架構(Object-Oriented Software Construction)探討

This paper presents an overview of progress in the area of multiple input multiple output (MIMO) space-time coded wireless systems. After some background on the research leading to the discovery of the enormous potential of MIMO wireless links, we highlight the different classes of techniques and algorithms proposed which attempt to realize the various benefits of MIMO including spatial multiplexing and space-time coding schemes. These algorithms are often derived and analyzed under ideal independent fading conditions. We present the state of the art in channel modeling and measurements, leading to a better understanding of actual MIMO gains. Finally, the paper addresses current questions regarding the integration of MIMO links in practical wireless systems and standards.

/pdf/from-theory-to-practice-an-overview-of-mimo-space-time-coded-431vesj02q.pdf

From theory to practice: an overview of MIMO space-time coded wireless systems

Silicon implementation of iterative detection and decoding for multi-antenna receivers

Summary form only given. Advanced communication systems obey a generalized Moore's law. Not only does hardware complexity double every 18 months, but also the other performance indicators such as program size or memory content-increase by a factor of two in a period of one and a half and three years, to mention two examples. The drawing force behind this growth is the algorithmic complexity which is needed to design communication systems operating close to the information theoretic limits: near optimum system performance is bound to exponentially increasing algorithmic complexity. Stated differently, the usefulness for the user only grows logarithmically with complexity. Basically, this logarithmic complexity provides the rational for the continued growth of the semiconductor industry. Advanced communication systems will be implemented as reconfigurable, heterogeneous multiprocessor platforms. This hypothesis is based on the fundamental trade-off between computational efficiency (MOPS/mW) and flexibility. While programmable devices (processors or DSPs) have the highest degree of flexibility, they have at least a two to three orders of magnitude smaller computational efficiency than the intrinsic computational efficiency (ICE) of fixed architectures. Hence, since power is the limiting factor, the SOCs of the future will carefully match algorithm with architecture to achieve an optimum. ("just as much flexibility as needed"). These SOC's will, therefore, become application specific platforms.

Why we need all these MIPS in future wireless communication systems-and how to design algorithms and architecture for these systems

Publisher Summary A paradigm change in designing complex systems-on-chip (SoCs) occurs roughly every 12 years because of the exponentially increasing number of transistors on a chip. This paradigm change is characterized by a move to a higher level of abstraction. Instead of thinking in register-transfer level (RTL) blocks and wires, computing elements and interconnect are needed to be thought. The next design discontinuity will lead to different solutions, depending on the application. The following core propositions for wireless communications are made: future SoC for wireless communications will be heterogeneous, reconfigurable Multi-Processor System-on-Chip (MPSoC). They will contain computational elements that cover the entire spectrum, from fixed functionality blocks to domain-specific DSPs and general-purpose processors. A key role will be played by ASIPs. ASIPs exploit the full architectural space (memory, interconnect, instruction set, parallelism), so they are optimally matched to a specific task. The heterogeneous computational elements will communicate via a network-on-chip (NoC), as the conventional bus structures do not scale. These MPSoC platforms will be designed by a cross-disciplinary team. This chapter substantiates this proposition. It begins by analyzing the properties of future wireless communication systems and observes that the systems are computationally demanding. Furthermore, they need innovative architectural concepts to be energy efficient. The chapter discusses the canonical structure of a digital receiver for wireless communication and addresses the design of ASIPs.

Opportunities for Application-Specific Processors: The Case of Wireless Communications

A fast simulation technique for the analysis of the system performance of digital receivers is presented and applied to the UMTS terrestrial radio access (UTRA). This semianalytical method comprises the computation of receiver output statistics, conditioned on a static channel model, a given spreading sequence and a given transmitted symbol sequence. Averaging over the known channel tap probability density function and over all symbol sequences yields the system performance, conditioned on the selected spreading sequence. This method permits the qualitative classification and comparison of different detection algorithms for WCDMA in the performance-complexity design space. In the UTRA environment, the performance is shown to vary significantly with the choice of the OVSF spreading sequence. Simulation results for a RAKE receiver in indoor and outdoor scenarios are compared to lower and upper bounds of the probability of error of a maximum likelihood sequence estimation (MLSE) algorithm. It is shown that a RAKE receiver provides sufficient detection performance in medium-rate outdoor scenarios, whereas in high-rate indoor scenarios, especially in the uplink, alternative receiver structures must be considered.

Exploring the UMTS WCDMA-receiver design space using a semianalytical approach

From an implementation as well as theoretical point of view it is a fundamental property of the symbol-by-symbol MAP decoding algorithm, that the dependence of the decoder output on the decoder inputs decays with distance in the code trellis. By using results from the theory of products of random matrices we present a general prove of this property and show that the rate of decay is exponentially with distance along the trellis. Furthermore, we examine how the rate of decay depends on the channel parameter and the a priori information, and how it evolves during the iterative decoding process. Finally, we comment on possible practical implications.

Heinrich Meyr

Papers

Silicon implementation of iterative detection and decoding for multi-antenna receivers

Why we need all these MIPS in future wireless communication systems-and how to design algorithms and architecture for these systems

Opportunities for Application-Specific Processors: The Case of Wireless Communications

Exploring the UMTS WCDMA-receiver design space using a semianalytical approach

Products of Random Matrices in Iterative (Turbo) Decoding