A decision forest system and decision forest reasoning method based on fpga

Abstract

The invention discloses a decision forest system and a decision forest reasoning method based on FPGA, belonging to the field of machine learning. The present invention realizes the decision tree on the FPGA based on the architecture of the path backtracking module and the classification module, which is conducive to automatically converting the trained decision tree MATLAB code into Verilog code without manually implementing a large number of decision trees in the EDA tool. Using a top-down design structure, each decision tree is run in parallel as a sub-module, which ensures accuracy while rapidly deploying, and consumes fewer hardware resources. Based on the pipeline technology, the majority voting method is implemented on the FPGA, and the decision tree sub-module is called in the top-level module and the results of the sub-modules are processed uniformly, so as to improve the reasoning speed. The asynchronous FIFO module transmits data across clock domains, making it more widely used. The reasoning realization method provided by the present invention is applicable to the decision tree generated by any training algorithm in principle.

Description

A Decision Forest System and Decision Forest Reasoning Method Based on FPGA technical field The invention belongs to the technical field of machine learning, more specifically, relate to a kind of decision forest system based on FPGA realization System and decision forest reasoning methods. Background technique [0002] Ensemble learning is an important method in machine learning. Ensemble learning combines multiple learners combined, often achieve significantly better generalization performance than a single learner. The whole ensemble learning does not depend on any single model, but Make predictions collectively. Decision forest is one of the most famous representatives of parallel ensemble learning methods, and its diversity is achieved by using obtained from different training data subsets. Decision forest is a decision tree based learner and a majority voting method as an ensemble method. An ensemble learning method composed of the formula can also further ...

AI Technical Summary

Problems solved by technology

In the prior art, a variety of algorithms used in the decision forest training process are disclosed, but the implementation method of reasoning in FPGA is still immature and needs to be further improved

Images