Exploiting Symmetric Temporally Sparse BPTT for Efficient RNN Training

Recurrent Neural Networks (RNNs) are useful in temporal sequence tasks. However, training RNNs involves dense matrix multiplications which require hardware that can support a large number of arithmetic operations and memory accesses. Implementing online training of RNNs on the edge calls for optimized algorithms for an efficient deployment on...

Analysis of Injury Severity of Drivers Involved Different Types of Two-Vehicle Crashes Using Random-Parameters Logit Models with Heterogeneity in Means and Variances

This study proposes random-parameters multinomial logit models, with heterogeneity in means and variances, to explore the differences in the factors influencing injury severities of drivers involved in different types of two-vehicle crashes. The models are verified using crash data from the United Kingdom (UK) over three years (2016–2018). Three...

A Self-Bias-Flip with Charge Recycle Interface Circuit with No External Energy Reservoir for Piezoelectric Energy Harvesting Array

This article presents a piezoelectric energy harvesting (PEH) interface circuit using a new self-bias-flip with the charge recycle (SBFR) technique without employing any additional energy reservoir. Traditional designs, including synchronous-switch harvesting on inductor (SSHI), synchronous-switch harvesting on capacitor (SSHC), synchronous...

Current and future trends in topology optimization for additive manufacturing

Manufacturing-oriented topology optimization has been extensively studied the past two decades, in particular for the conventional manufacturing methods, for example, machining and injection molding or casting. Both design and manufacturing engineers have benefited from these efforts because of the close-to-optimal and friendly-to-manufacture...