Simultaneous direct measurement of the electrocaloric and dielectric dynamics of ferroelectrics with microsecond temporal resolution
Authors:
J. Fischer,
J. Döntgen,
C. Molin,
S. E. Gebhardt,
Y. Hambal,
V. V. Shvartsman,
D. C. Lupascu,
D. Hägele,
J. Rudolph
Abstract:
A contactless technique for direct time-resolved measurements of the full dynamics of the adiabatic temperature change in electrocaloric materials is introduced. The infrared radiation emitted by the electrocaloric sample is sensitively detected with $μ$s time resolution and mK temperature resolution. We present time-resolved measurements of the electrocaloric effect up to kHz frequencies of the d…
▽ More
A contactless technique for direct time-resolved measurements of the full dynamics of the adiabatic temperature change in electrocaloric materials is introduced. The infrared radiation emitted by the electrocaloric sample is sensitively detected with $μ$s time resolution and mK temperature resolution. We present time-resolved measurements of the electrocaloric effect up to kHz frequencies of the driving electric field and down to small field strengths. The simultaneous recording of transients for applied electric field and induced polarization gives a comprehensive view on the correlation of electrocaloric and ferroelectric properties. The technique can further be applied to the continuous measurement of fatigue for $> 10^6$ electric field cycles.
△ Less
Submitted 18 April, 2023; v1 submitted 5 February, 2023;
originally announced February 2023.
Direct Feedback Alignment with Sparse Connections for Local Learning
Authors:
Brian Crafton,
Abhinav Parihar,
Evan Gebhardt,
Arijit Raychowdhury
Abstract:
Recent advances in deep neural networks (DNNs) owe their success to training algorithms that use backpropagation and gradient-descent. Backpropagation, while highly effective on von Neumann architectures, becomes inefficient when scaling to large networks. Commonly referred to as the weight transport problem, each neuron's dependence on the weights and errors located deeper in the network require…
▽ More
Recent advances in deep neural networks (DNNs) owe their success to training algorithms that use backpropagation and gradient-descent. Backpropagation, while highly effective on von Neumann architectures, becomes inefficient when scaling to large networks. Commonly referred to as the weight transport problem, each neuron's dependence on the weights and errors located deeper in the network require exhaustive data movement which presents a key problem in enhancing the performance and energy-efficiency of machine-learning hardware. In this work, we propose a bio-plausible alternative to backpropagation drawing from advances in feedback alignment algorithms in which the error computation at a single synapse reduces to the product of three scalar values. Using a sparse feedback matrix, we show that a neuron needs only a fraction of the information previously used by the feedback alignment algorithms. Consequently, memory and compute can be partitioned and distributed whichever way produces the most efficient forward pass so long as a single error can be delivered to each neuron. Our results show orders of magnitude improvement in data movement and $2\times$ improvement in multiply-and-accumulate operations over backpropagation. Like previous work, we observe that any variant of feedback alignment suffers significant losses in classification accuracy on deep convolutional neural networks. By transferring trained convolutional layers and training the fully connected layers using direct feedback alignment, we demonstrate that direct feedback alignment can obtain results competitive with backpropagation. Furthermore, we observe that using an extremely sparse feedback matrix, rather than a dense one, results in a small accuracy drop while yielding hardware advantages. All the code and results are available under https://github.com/bcrafton/ssdfa.
△ Less
Submitted 9 May, 2019; v1 submitted 30 January, 2019;
originally announced March 2019.
