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

The availability of C compilers is crucial to the efficient design of embedded systems. Using virtual resources to automatically generate parts of a compiler's instruction scheduler from a formal processor description significantly reduces the overall scheduler generation time.

Instruction scheduler generation for retargetable compilation

The evolving digital television broadcasting standard does not standardize the data rate of the transmitted data but instead leaves it completely unspecified. We propose two digital receiver algorithms for the processing of an extended range of variable sample rates. The algorithms are compared in terms of their complexity and performance. One of the algorithms is based on filtering the received samples prior to timing synchronization, whereas the second algorithm increases the sample rate in the timing recovery loop and matched filter. Both algorithms can be implemented causing negligible loss in a DVB receiver.

http://ieeexplore.ieee.org/iel3/4477/12726/00587666.pdf

Variable sample rate digital feedback NDA timing synchronization

Phase locked loops (PLL) for RF carrier synthesis often employ oscillators that insert a considerable amount of time varying phase noise into the received signal. That noise must then be removed in the digital baseband receiver. This phase noise is an indivisible superposition of noise components from receiver and transmitter. Regarding systems with multiple transmit and receive antennas (MIMO) and if multiple PLL for carrier synthesis are used each of the superposed phase noise processes per transmit and receive antenna pair can be measured at the receiver. This paper provides a new scheme for high SNR scenarios that exploits spatial correlation between these overlaying phase noise processes at the receiver in order to improve estimation and compensation of the phase noise. Therefore the Wiener filter approach is applied.

Improving MIMO phase noise estimation by exploiting spatial correlations

The design of digital signal processing systems typically consists of an algorithm development phase carried out at a behavioral level and the selection of an efficient hardware architecture for implementation. In order to speed up the joint optimization of algorithms and architectures, a fast path to implementation must be provided. This can be achieved efficiently by directly mapping the data flow specification of the system to an RTL target architecture by means of HDL code generation. For algorithm design, communication systems are most easily modeled using multirate data flow graphs in which no notion of time is maintained. HDL code generation introduces a cycle based timing model and maps the data flow models to RTL implementations, which are usually taken from a library. Due to the increase in ASIC design complexity, these building blocks reach a high level of functionality and have complex interfacing properties. Therefore, it becomes necessary to generate additional interfacing and controlling hardware to synthesize an operable system. In this paper, we present a new approach of mapping multirate dataflow graphs to complex RTL hardware models and derive algorithms to synthesize these high-level RTL building blocks into a complete operable system.

Mapping multirate dataflow to complex RT level hardware models

This article is an introduction to the FRIDGE design environment which supports the design and DSP implementation of fixed-point digital signal processing systems. We present the tool-supported transformation of signal processing algorithms coded in floating-point ANSI C to a fixed-point representation in SystemC. We introduce the novel approach to control and data flow analysis, which is necessary for the transformation. The design environment enables fast bit-true simulation by mapping the fixed-point algorithm to integral data types of the host machine. A speedup by a factor of 20 to 400 can be achieved compared to C++-library-based bit-true simulation. FRIDGE also provides a direct link to DSP implementation by processor specific C code generation and advanced code optimization.

/pdf/design-and-dsp-implementation-of-fixed-point-systems-303ihp5qpm.pdf

Heinrich Meyr

Papers

Instruction scheduler generation for retargetable compilation

Variable sample rate digital feedback NDA timing synchronization

Improving MIMO phase noise estimation by exploiting spatial correlations

Mapping multirate dataflow to complex RT level hardware models

Design and DSP implementation of fixed-point systems