Deep Reinforcement Learning (DRL) is considered a potential framework to improve many real-world autonomous systems; it has attracted the attention of multiple and diverse fields. Nevertheless, the successful deployment in the real world is a test most of DRL models still need to pass. In this work we focus on this issue by reviewing and evaluating the research efforts from both domain-agnostic and domain-specific communities. On one hand, we offer a comprehensive summary of DRL challenges and summarize the different proposals to mitigate them; this helps identifying five gaps of domain-agnostic research. On the other hand, from the domain-specific perspective, we discuss different success stories and argue why other models might fail to be deployed. Finally, we take up on ways to move forward accounting for both perspectives.

Most approaches in reinforcement learning (RL) are data-hungry and specific to fixed environments. In this paper, we propose a principled framework for adaptive RL, called AdaRL, that adapts reliably to changes across domains. Specifically, we construct a generative environment model for the structural relationships among variables in the system and embed the changes in a compact way, which provides a clear and interpretable picture for locating what and where the changes are and how to adapt. Based on the environment model, we characterize a minimal set of representations, including both domain-specific factors and domain-shared state representations, that suffice for reliable and low-cost transfer. Moreover, we show that by explicitly leveraging a compact representation to encode changes, we can adapt the policy with only a few samples without further policy optimization in the target domain. We illustrate the efficacy of AdaRL through a series of experiments that allow for changes in different components of Cartpole and Atari games.

The square kernel is a standard unit for contemporary Convolutional Neural Networks (CNNs), as it fits well on the tensor computation for the convolution operation. However, the receptive field in the human visual system is actually isotropic like a circle. Motivated by this observation, we propose using circle kernels with isotropic receptive fields for the convolution, and our training takes approximately equivalent amount of calculation when compared with the corresponding CNN with square kernels. Our preliminary experiments demonstrate the rationality of circle kernels. We then propose a kernel boosting strategy that integrates the circle kernels with square kernels for the training and inference, and we further let the kernel size/radius be learnable during the training. Note that we reparameterize the circle kernels or integrated kernels before the inference, thus taking no extra computation as well as the number of parameter overhead for the testing. Extensive experiments on several standard datasets, ImageNet, CIFAR-10 and CIFAR-100, using the circle kernels or integrated kernels on typical existing CNNs, show that our approach exhibits highly competitive performance. Specifically, on ImageNet with standard data augmentation, our approach dramatically boosts the performance of MobileNetV3-Small by 5.20% top-1 accuracy and 3.39% top-5 accuracy, and boosts the performance of MobileNetV3-Large by 2.16% top-1 accuracy and 1.18% top-5 accuracy. The first three authors contribute equally.

Deep neural networks will move past their shortcomings without help from symbolic artificial intelligence, three pioneers of deep learning argue in a paper published in the July issue of the Communications of the ACM journal. In their paper, Yoshua Bengio, Geoffrey Hinton, and Yann LeCun, recipients of the 2018 Turing Award, explain the current challenges of deep learning and how it differs from learning in humans and animals. They also explore recent advances in the field that might provide blueprints for the future directions for research in deep learning. Titled "Deep Learning for AI," the paper envisions a future in which deep learning models can learn with little or no help from humans, are flexible to changes in their environment, and can solve a wide range of reflexive and cognitive problems. Deep learning is often compared to the brains of humans and animals.

Video games have, for several decades, only loaded the assets they needed to use, as they used them, due to tiny RAM amounts relative to the asset sizes yea. The original Playstation is a good example with its 2MB of RAM and 800MB discs full of assets. Enabling the ability to do something at all for many, is often more important than being able to do it extremely quickly for a few. Modern games get around this in part, by having the graphics driver handle all the asset management in and out of VRAM, allowing the driver to swap the least recently used assets out to RAM until they are needed, and then swap them back to VRAM, without the executing program having to know it happened. Basically treating all of VRAM and virtual VRAM hosted in RAM as one big asymmetrically speedy storage medium.

Deep neural networks will move past their shortcomings without help from symbolic artificial intelligence, three pioneers of deep learning argue in a paper published in the July issue of the Communications of the ACM journal. In their paper, Yoshua Bengio, Geoffrey Hinton, and Yann LeCun, recipients of the 2018 Turing Award, explain the current challenges of deep learning and how it differs from learning in humans and animals. They also explore recent advances in the field that might provide blueprints for the future directions for research in deep learning. Titled "Deep Learning for AI," the paper envisions a future in which deep learning models can learn with little or no help from humans, are flexible to changes in their environment, and can solve a wide range of reflexive and cognitive problems. Deep learning is often compared to the brains of humans and animals.

A new deep learning tool could help slash the time it takes to interpret cardiology scans to diagnose obstructive coronary artery disease by 75 percent. The study, featured in the European Journal of Radiology, describes how this new technology, powered by artificial intelligence (AI) can not only accelerate processing times but also assist radiologists in identifying potential problem areas in the patients' major arteries. In obstructive coronary arterial disease, the blood vessels that supply the heart gradually narrow, and in severe classes, can even close completely. This happens as a result of the accumulation of plaque in a process called atherosclerosis. Checking for obstructions and blockages in coronary scans can be challenging--radiologists require specialized training and extensive experience in order to be able to interpret these scans correctly.

Applications are invited for a new prestigious DeepMind Academic Fellow in Machine Learning at Queen Mary University of London. Following a recent donation to the University from DeepMind, this three-year Fellowship is created to provide an opportunity for an excellent early career researcher in the fields of Computer Science and/or Machine Learning/ Artificial Intelligence to further their research and prepare for a full academic role within a supportive environment. The ideal candidate will have completed a PhD in a relevant field (or expect to have completed by this September) and have clear and ambitious plans for their future research, alongside the enthusiasm to act as a role model for Black researchers of the future. We particularly encourage applications from those who are in under-represented groups, and particularly those who identify as Black, as Black staff are under-represented at this level within the School of Electronic Engineering and Computer Science at Queen Mary. The Fellowship will be research-focused and the successful candidate will be allocated a research studentship to support outputs.

So you have been doing some deep learning, training some models using TensorFlow, PyTorch, or whatever library you are fond of. You feel like you are getting a grip on this thing and think it can turn out as a possible career option. Then comes the first professional assignment, it could be a freelance project you take up or something your company assigns to you and boom you feel like a person on a raft out in the sea, and nothing to guide you. Well, when working with datasets of the self-created origin or something that isn't a part of the precreated dataset pipelines created out there then there are several problems you may face. How do I clean this dataset?

Increasing number of parameters, latency, resources required to train etc have made working with deep learning tricky. Google researchers, in an extensive survey, have found common challenging areas for deep learning practitioners and suggested key checkpoints to mitigate these challenges. According to Gaurav Menghani of Google Research, if one were to deploy a model on smartphones where inference is constrained or expensive due to cloud servers, attention should be paid to inference efficiency. And if a large model has to be trained from scratch with limited training resources, models that are designed for training efficiency would be better off. According to Menghani, practitioners should aim to achieve pareto-optimality i.e. any model we choose should have the best of tradeoffs.