Search for: All records

This paper presents a novel technique to reduce energy consumption of a machine learning classifier based on incremental-precision feature computation and classification. Specifically, the algorithm starts with features computed using the lowest possible precision. Depending on the classification accuracy, the features of the previous level are combined with features of the incremental-precision to compute the features in higher-precision. This process is continued till a desired accuracy is obtained. A certain threshold that allows many samples to be classified using a low-precision classifier can reduce energy consumption, but increases misclassification error. To implement hardware which provides the required updates in precision, an incremental-precision architecture based on data-path decomposition is proposed. One novel aspect of this work lies in the design of appropriate thresholds for multi-level classification using training data such that a family of designs can be obtained that enable trade-offs between classification accuracy and energy consumption. Another novel aspect involves the design of hardware architectures based on data-path decomposition which can incrementally increase precision upon demand. Using a seizure detection example, it is shown that the proposed incremental-precision based multi-level classification approach can reduce energy consumption by 35% while maintaining high sensitivity, or by about 50% at the expense of 15% degradation in sensitivity compared to similar approaches to seizure detection in literature. The reduction in energy is achieved at the expense of small area, timing and memory overheads as multiple classification steps are used instead of a single step. 

This paper presents a novel incremental-precision classification approach that leads to a reduction in energy consumption of linear classifiers for IoT applications. Features are first input to a low-precision classifier. If the classifier successfully classifies the sample, then the process terminates. Otherwise, the classification performance is incrementally improved by using a classifier of higher precision. This process is repeated until the classification is complete. The argument is that many samples can be classified using the low-precision classifier, leading to a reduction in energy. To achieve incremental-precision, a novel data-path decomposition is proposed to design of fixed-width adders and multipliers. These components improve the precision without recalculating the outputs, thus reducing energy. Using a linear classification example, it is shown that the proposed incremental-precision based multi-level classifier approach can reduce energy by about 41% while achieving comparable accuracies as that of a full-precision system.