Therefore, we need to specialize the feature extractor to best match the target dataset. The peak memory cost of this phase is 61MB under resolution 224, which is reached when the largest sub-network is sampled. MAC (only forward) of sampled sub-nets is (355M + 1182M) / 2 = 768.5M Therefore, the total MAC of this phase is 768.5M Flowers, where 2040 is the number of total training samples, 0.2 means the validation set consists of Details of the accuracy predictor is provided in Appendix B. It takes the one-hot encoding of the sub-network's MAC of this accuracy predictor is only 0.37M, which is 3-4 orders of magnitude smaller than the Therefore, TinyTL is not only more memory-efficient but also more computation-efficient.

We present APQ for efficient deep learning inference on resource-constrained hardware. Unlike previous methods that separately search the neural architecture, pruning policy, and quantization policy, we optimize them in a joint manner. To deal with the larger design space it brings, a promising approach is to train a quantization-aware accuracy predictor to quickly get the accuracy of the quantized model and feed it to the search engine to select the best fit. However, training this quantization-aware accuracy predictor requires collecting a large number of quantized pairs, which involves quantization-aware finetuning and thus is highly time-consuming. To tackle this challenge, we propose to transfer the knowledge from a full-precision (i.e., fp32) accuracy predictor to the quantization-aware (i.e., int8) accuracy predictor, which greatly improves the sample efficiency. Besides, collecting the dataset for the fp32 accuracy predictor only requires to evaluate neural networks without any training cost by sampling from a pretrained once-for-all network, which is highly efficient. Extensive experiments on ImageNet demonstrate the benefits of our joint optimization approach. With the same accuracy, APQ reduces the latency/energy by 2x/1.3x over MobileNetV2+HAQ. Compared to the separate optimization approach (ProxylessNAS+AMC+HAQ), APQ achieves 2.3% higher ImageNet accuracy while reducing orders of magnitude GPU hours and CO2 emission, pushing the frontier for green AI that is environmental-friendly. The code and video are publicly available.

"exploring the humanizing of AI by building a digital brain which can be used as a platform for autonomously animating hyper-realistic digital humans" "I think what will be increasingly important in the digital human space is ethics, as they relate both to the digital human and to the real-life people who may be impacted. From a digital human perspective, companies are essentially birthing entities which, in many cases, are expected to form meaningful connections and relationships with people. So how organizations treat these digital humans--including any decision to dispose of them if they are no longer deemed needed--will increasingly become important. On the flipside, entertainment organizations that are using digital humans run the risk of causing concern of replacing real humans […] and it will be important to clarify how and why digital humans are being used in lieu of the'real' thing." Excerpts from this article: The Virtual Beings Are Arriving Efficient deployment of deep learning models requires specialized neural network architectures to best fit different hardware platforms and efficiency constraints (defined as deployment scenarios).

Efficient deployment of deep learning models requires specialized neural network architectures to best fit different hardware platforms and efficiency constraints (defined as deployment scenarios). Traditional approaches either manually design or use AutoML to search a specialized neural network and train it from scratch for each case. It is expensive and unscalable since their training cost is linear w.r.t. the number of deployment scenarios. In this work, we introduce Once for All (OFA) for efficient neural network design to handle many deployment scenarios, a new methodology that decouples model training from architecture search. Instead of training a specialized model for each case, we propose to train a once-for-all network that supports diverse architectural settings (depth, width, kernel size, and resolution). Given a deployment scenario, we can later search a specialized sub-network by selecting from the once-for-all network without training. As such, the training cost of specialized models is reduced from O(N) to O(1). However, it's challenging to prevent interference between many sub-networks. Therefore we propose the progressive shrinking algorithm, which is capable of training a once-for-all network to support more than $10^{19}$ sub-networks while maintaining the same accuracy as independently trained networks, saving the non-recurring engineering (NRE) cost. Extensive experiments on various hardware platforms (Mobile/CPU/GPU) and efficiency constraints show that OFA consistently achieves the same level (or better) ImageNet accuracy than SOTA neural architecture search (NAS) methods. Remarkably, OFA is orders of magnitude faster than NAS in handling multiple deployment scenarios (N). With N=40, OFA requires 14x fewer GPU hours than ProxylessNAS, 16x fewer GPU hours than FBNet and 1,142x fewer GPU hours than MnasNet. The more deployment scenarios, the more savings over NAS.