Reconfigurable multi-class support vector machine system

Abstract

The invention relates to a reconfigurable multi-class support vector machine system. The system comprises a main control module, a storage control module, a kernel function operation module, a class operation module and a result comparison module. The main control module provides control information and reconstruction information for the whole decision-making process; the storage control module isused for controlling the storage of data; the kernel function operation module calculates a kernel function between the test data and the support vector; the category operation module is used for calculating a decision value and a classification category; and the result comparison module compares decision values calculated by different models to obtain a final classification result of the test data. Compared with a traditional method, the parallelism of hardware is fully utilized, the operation speed of support vector machine classification is increased, the kernel function operation module and the category operation module share calculation resources, hardware reconfiguration is supported, and good flexibility is achieved for samples with different feature numbers.

Description

technical field [0001] The invention belongs to the field of hardware realization of machine learning algorithms, in particular to a reconfigurable multi-classification support vector machine system. Background technique [0002] Support vector machine (SVM) is a supervised machine learning algorithm that can be used for data analysis, pattern recognition, data classification and regression analysis, and is widely used. The support vector machine algorithm can be divided into two parts: training and decision-making. The training part is generally performed on a high-performance server due to the complex calculation method and large amount of calculation. The decision-making part uses the trained model to classify or regression analysis the sample data. Its calculation process includes a large number of multiplication and addition operations. Note that the input sample is x, and the i-th support vector is x i , and its corresponding coefficient is α i , and its correspondi...

AI Technical Summary

Problems solved by technology

There are a large number of Cache and control units in the CPU, and the number of ALUs used for calculation is small, while the GPU has a large number of computing resources and storage resources, and the control is simple. Therefore, for calculation-intensive algorithms, the GPU has a faster calculation speed than the CPU. However, the GPU has a large area and high energy consumption, so it is not suitable for embedded real-time computing

Images