AttriNet

Abstract

Human activity recognition (HAR) is essential to many contextaware applications in mobile and ubiquitous computing. A human's physical activity can be decomposed into a sequence of simple actions or body movements, corresponding to what we denote as mid-level features. Such mid-level features ("leg up, " 'leg down, " "leg still, " ...), which we contrast to high-level activities ("walking, " "sitting, " ...) and low-level features (raw sensor readings), can be developed manually. While proven to be effective, this manual approach is not scalable and relies heavily on human domain expertise. In this paper, we address this limitation by proposing a machine learning method, At-triNet, based on deep belief networks. Our AttriNet method automatically constructs mid-level features and outperforms baseline approaches. Interestingly, we show in experiments that some of the features learned by AttriNet highly correlate with manually defined features. This result demonstrates the potential of using deep learning techniques for learning mid-level features that are semantically meaningful, as a replacement to handcrafted features. Generally, this empirical finding provides an improved understanding of deep learning methods for HAR. CCS CONCEPTS • Computing methodologies → Artificial intelligence; Activity recognition and understanding.

Cortical Columns Computing Systems: Microarchitecture Model, Functional Building Blocks, and Design Tools

“…As illustrated in Figure 5, this framework consists of three key components: (1) a microarchitecture model that can facilitate RTL implementation of Cortical Columns (CCs) and Reference Frames (RFs) employed in C3S designs; (2) a suite of highly optimized functional units and building blocks implemented in System Verilog to support efficient application-specific implementations of C3S designs; and (3) a software tool suite consisting of a PyTorch simulator for rapid design space exploration of C3S designs and a design synthesis flow for mapping PyTorch functional models to corresponding C3S hardware. Specific applications of current interest include visual object recognition, anomaly detection on time-series signals (e.g., ECG), edge-native always-on keyword spotting, and multi-modal human activity recognition (HAR) [56].…”

“…Figure 5.Proposed framework for implementing C3S designs consists of three key components: Microarchitecture model[11], functional building blocks[12] and design exploration tools. Some applications of interest are visual object recognition, anomaly detection[13], keyword spotting, and human activity recognition[56]. These framework components currently support TNN design and implementation and key relevant publications are cited here.…”