Search for: All records

Abstract Energy efficiency in computation is ultimately limited by noise, with quantum limits setting the fundamental noise floor. Analog physical neural networks hold promise for improved energy efficiency compared to digital electronic neural networks. However, they are typically operated in a relatively high-power regime so that the signal-to-noise ratio (SNR) is large (>10), and the noise can be treated as a perturbation. We study optical neural networks where all layers except the last are operated in the limit that each neuron can be activated by just a single photon, and as a result the noise on neuron activations is no longer merely perturbative. We show that by using a physics-based probabilistic model of the neuron activations in training, it is possible to perform accurate machine-learning inference in spite of the extremely high shot noise (SNR  ~ 1). We experimentally demonstrated MNIST handwritten-digit classification with a test accuracy of 98% using an optical neural network with a hidden layer operating in the single-photon regime; the optical energy used to perform the classification corresponds to just 0.038 photons per multiply-accumulate (MAC) operation. Our physics-aware stochastic training approach might also prove useful with non-optical ultra-low-power hardware. 

Abstract BackgroundPrioritizing wild relative diversity for improving crop adaptation to emerging drought-prone environments is challenging. Here, we combine the genome-wide environmental scans (GWES) in wheat diploid ancestorAegilops tauschii(Ae. tauschii) with allele testing in the genetic backgrounds of adapted cultivars to identify diversity for improving wheat adaptation to water-limiting conditions. ResultsWe evaluate the adaptive allele effects inAe. tauschii-wheat introgression lines phenotyped for multiple traits under irrigated and water-limiting conditions using both unmanned aerial system-based imaging and conventional approaches. The GWES show that climatic gradients alone explain more than half of genomic variation inAe. tauschii, with many alleles associated with climatic factors inAe. tauschiibeing linked with improved performance of introgression lines under water-limiting conditions. We find that the most significant GWES signals associated with temperature annual range in the wild relative are linked with reduced canopy temperature in introgression lines and increased yield. ConclusionsOur results suggest that introgression of climate-adaptive alleles fromAe. tauschiihas the potential to improve wheat performance under water-limiting conditions, and that variants controlling physiological processes responsible for maintaining leaf temperature are likely among the targets of adaptive selection in a wild relative. Adaptive variation uncovered by GWES in wild relatives has the potential to improve climate resilience of crop varieties.