Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

/pdf/deep-residual-learning-for-image-recognition-b75jg61qrq.pdf

Deep Residual Learning for Image Recognition

In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. 
The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Deep learning is a form of machine learning that enables computers to learn from experience and understand the world in terms of a hierarchy of concepts. Because the computer gathers knowledge from experience, there is no need for a human computer operator to formally specify all the knowledge that the computer needs. The hierarchy of concepts allows the computer to learn complicated concepts by building them out of simpler ones; a graph of these hierarchies would be many layers deep. This book introduces a broad range of topics in deep learning. The text offers mathematical and conceptual background, covering relevant concepts in linear algebra, probability theory and information theory, numerical computation, and machine learning. It describes deep learning techniques used by practitioners in industry, including deep feedforward networks, regularization, optimization algorithms, convolutional networks, sequence modeling, and practical methodology; and it surveys such applications as natural language processing, speech recognition, computer vision, online recommendation systems, bioinformatics, and videogames. Finally, the book offers research perspectives, covering such theoretical topics as linear factor models, autoencoders, representation learning, structured probabilistic models, Monte Carlo methods, the partition function, approximate inference, and deep generative models. Deep Learning can be used by undergraduate or graduate students planning careers in either industry or research, and by software engineers who want to begin using deep learning in their products or platforms. A website offers supplementary material for both readers and instructors.

Deep Learning

State-of-the-art exoskeletons are typically limited by the low control bandwidth and small-range stiffness of actuators, which are based on high gear ratios and elastic components (e.g., series elastic actuators). Furthermore, most exoskeletons are based on discrete gait phase detection and/or discrete stiffness control, resulting in discontinuous torque profiles. To fill these two gaps, we developed a portable, lightweight knee exoskeleton using quasi-direct-drive (QDD) actuation that provides 14 N·m torque (36.8% biological joint moment for overground walking). This article presents 1) stiffness modeling of torque-controlled QDD exoskeletons and 2) stiffness-based continuous torque controller that estimates knee joint moment in real-time. Experimental tests found that the exoskeleton had a high bandwidth of stiffness control (16 Hz under 100 N·m/rad) and high torque tracking accuracy with 0.34 N·m root mean square error (6.22%) across 0–350 N·m/rad large-range stiffness. The continuous controller was able to estimate knee moments accurately and smoothly for three walking speeds and their transitions. Experimental results with eight able-bodied subjects demonstrated that our exoskeleton was able to reduce the muscle activities of all eight measured knee and ankle muscles by 8.60%–15.22% relative to the unpowered condition and two knee flexors and one ankle plantar flexor by 1.92%–10.24% relative to the baseline (no exoskeleton) condition.

Modeling and Stiffness-Based Continuous Torque Control of Lightweight Quasi-Direct-Drive Knee Exoskeletons for Versatile Walking Assistance

Real-Time Robust 3D Plane Extraction for Wearable Robot Perception and Control

Quick Response (QR) code is one of the most worldwide used two-dimensional codes.~Traditional QR codes appear as random collections of black-and-white modules that lack visual semantics and aesthetic elements, which inspires the recent works to beautify the appearances of QR codes. However, these works adopt fixed generation algorithms and therefore can only generate QR codes with a pre-defined style. In this paper, combining the Neural Style Transfer technique, we propose a novel end-to-end method, named ArtCoder, to generate the stylized QR codes that are personalized, diverse, attractive, and scanning-robust.~To guarantee that the generated stylized QR codes are still scanning-robust, we propose a Sampling-Simulation layer, a module-based code loss, and a competition mechanism. The experimental results show that our stylized QR codes have high-quality in both the visual effect and the scanning-robustness, and they are able to support the real-world application.

An End-to-end Method for Producing Scanning-robust Stylized QR Codes.

Recently, deep learning-based denoising approaches have led to dramatic improvements in low sample-count Monte Carlo rendering. These approaches are aimed at path tracing, which is not ideal for simulating challenging light transport effects like caustics, where photon mapping is the method of choice. However, photon mapping requires very large numbers of traced photons to achieve high-quality reconstructions. In this paper, we develop the first deep learning-based method for particle-based rendering, and specifically focus on photon density estimation, the core of all particle-based methods. We train a novel deep neural network to predict a kernel function to aggregate photon contributions at shading points. Our network encodes individual photons into per-photon features, aggregates them in the neighborhood of a shading point to construct a photon local context vector, and infers a kernel function from the per-photon and photon local context features. This network is easy to incorporate in many previous photon mapping methods (by simply swapping the kernel density estimator) and can produce high-quality reconstructions of complex global illumination effects like caustics with an order of magnitude fewer photons compared to previous photon mapping methods.

Deep Photon Mapping.

Anaerobic digestion (AD) of biodegradable organics depends heavily diverse functional genera, but an attempt to simultaneously enhance the various AD phases including the rate-limiting step (hydrolysis acidification) and methanogenesis has often remained a challenge. Although zero-valent iron (ZVI) supplementation had been proposed by many, its effect on the biomethanation process and microbial community evolution has not been fully explored. In this study, the response of hydrolysis acidification, methanogenesis, kinetics, pollutant removal efficiency, and microbial community evolution to various doses of ZVI employed in AD of food waste leachate was investigated and reported. When 10 g/L ZVI was dosed in the AD system, hydrolysis acidification, COD removal, biomethanation, and methane percentage were elevated by 18.9%, 67%, 39%, and 36%, respectively, as opposed to the control reactor (R1). Kinetics (R2 of 0.998) revealed shorter lag phase (ƛ = 4.18 days) in ZVI-dosed reactors in contrast to longer periods (6.54 days) in R1. Methane production rate constant (k), sludge activity, and methane production rate estimated were 0.066 (day−1), 0.83 gCH4/gVSS, and 64.20 mLCH4/gVSS/day, respectively, suggesting ZVI promoted methane production rate. Hydrolytic and methanogenic phyla including Firmicutes, Chloroflexi, Bacteroidetes, and Euryarchaeota were more enriched in ZVI-dosed reactors as opposed to R1. Acetoclastic-(Methanosaeta) and hydrogenotrophic-(Methanobacterium) methanogens were the most abundant genera of the archaea within ZVI-dosed assays, indicating acetoclastic methanogenesis was the main pathway.

Hao Su

Papers

Modeling and Stiffness-Based Continuous Torque Control of Lightweight Quasi-Direct-Drive Knee Exoskeletons for Versatile Walking Assistance

Real-Time Robust 3D Plane Extraction for Wearable Robot Perception and Control

An End-to-end Method for Producing Scanning-robust Stylized QR Codes.

Deep Photon Mapping.

Response of hydrolysis, methanogenesis, and microbial community structure to iron dose during anaerobic digestion of food waste leachate