Stochastic Mixed-Signal VLSI Architecture for High-Dimensional Kernel Machines
نویسندگان
چکیده
A mixed-signal paradigm is presented for high-resolution parallel innerproduct computation in very high dimensions, suitable for efficient implementation of kernels in image processing. At the core of the externally digital architecture is a high-density, low-power analog array performing binary-binary partial matrix-vector multiplication. Full digital resolution is maintained even with low-resolution analog-to-digital conversion, owing to random statistics in the analog summation of binary products. A random modulation scheme produces near-Bernoulli statistics even for highly correlated inputs. The approach is validated with real image data, and with experimental results from a CID/DRAM analog array prototype in 0.5 m CMOS.
منابع مشابه
Massively Parallel Mixed-Signal VLSI Kernel Machines
Recently it has been shown that a simple learning paradigm, the support vector machine (SVM), outperforms some of the most elaborately tuned expert systems and neural networks in object recognition tasks. In run-time, the SVM operates by computing a kernelbased distance between the object’s vector at the input and a set of support vectors selected from the training set, and weighting the result...
متن کاملDynamic Circular Cellular Networks for Adaptive Smoothing of Multi-dimensional Signals
In 10] a theoretical framework for locally-adaptive smoothing of multi-dimensional data was presented. Based on this framework we introduce a hardware ef-cient architecture suitable for mixed-mode VLSI implementation. Substantial shortcomings of analogue implementations are overcome by connecting all cells in a circular structure: i) innuence of process parameter deviation ii) limited number of...
متن کاملDesign and Implementation of Digital Demodulator for Frequency Modulated CW Radar (RESEARCH NOTE)
Radar Signal Processing has been an interesting area of research for realization of programmable digital signal processor using VLSI design techniques. Digital Signal Processing (DSP) algorithms have been an integral design methodology for implementation of high speed application specific real-time systems especially for high resolution radar. CORDIC algorithm, in recent times, is turned out to...
متن کاملA One-Dimensional Analog VLSI Implementation for Nonlinear Real-Time Signal Preprocessing
R econstruction of given noisy data is an ill-posed problem and a computationally intensive task. Nonlinear regularization techniques are used to find a unique solution under certain constraints. In our contribution we present a parallel mixed-signal architecture which solves this nonlinear problem with in microseconds. By connecting all parallel cells in a circular manner it is possible to pro...
متن کاملA Scalable Vlsi Architecture Formultichannel Blind Deconvolutionand Source
In this paper, we describe a scalable VLSI architecture for a signal processing system that separates multiple independent source signals from a set of linear, convolved mixtures. The architecture employs a recently-proposed entropy-based algorithm and consists of a two-dimensional array of interconnected chips, each of which implements a two-input, two-output signal separation system with a ma...
متن کامل