As wireless networks evolve towards high mobility and providing better support for connected vehicles, a number of new challenges arise due to the resulting high dynamics in vehicular environments and thus motive rethinking of traditional wireless design methodologies. Future intelligent vehicles, which are at the heart of high mobility networks, are increasingly equipped with multiple advanced onboard sensors and keep generating large volumes of data. Machine learning, as an effective approach to artificial intelligence, can provide a rich set of tools to exploit such data for the benefit of the networks. In this article, we first identify the distinctive characteristics of high mobility vehicular networks and motivate the use of machine learning to address the resulting challenges. After a brief introduction of the major concepts of machine learning, we discuss its applications to learn the dynamics of vehicular networks and make informed decisions to optimize network performance. In particular, we discuss in greater detail the application of reinforcement learning in managing network resources as an alternative to the prevalent optimization approach. Finally, some open issues worth further investigation are highlighted.

In many environments only a tiny subset of all states yield high reward. In these cases, few of the interactions with the environment provide a relevant learning signal. Hence, we may want to preferentially train on those high-reward states and the probable trajectories leading to them. To this end, we advocate for the use of a backtracking model that predicts the preceding states that terminate at a given high-reward state. We can train a model which, starting from a high value state (or one that is estimated to have high value), predicts and sample for which the (state, action)-tuples may have led to that high value state. These traces of (state, action) pairs, which we refer to as Recall Traces, sampled from this backtracking model starting from a high value state, are informative as they terminate in good states, and hence we can use these traces to improve a policy. We provide a variational interpretation for this idea and a practical algorithm in which the backtracking model samples from an approximate posterior distribution over trajectories which lead to large rewards. Our method improves the sample efficiency of both on- and off-policy RL algorithms across several environments and tasks.

Deep learning continues to push state-of-the-art performance for the semantic segmentation of color (i.e., RGB) imagery; however, the lack of annotated data for many remote sensing sensors (i.e. hyperspectral imagery (HSI)) prevents researchers from taking advantage of this recent success. Since generating sensor specific datasets is time intensive and cost prohibitive, remote sensing researchers have embraced deep unsupervised feature extraction. Although these methods have pushed state-of-the-art performance on current HSI benchmarks, many of these tools are not readily accessible to many researchers. In this letter, we introduce a software pipeline, which we call EarthMapper, for the semantic segmentation of non-RGB remote sensing imagery. It includes self-taught spatial-spectral feature extraction, various standard and deep learning classifiers, and undirected graphical models for post-processing. We evaluated EarthMapper on the Indian Pines and Pavia University datasets and have released this code for public use.

Visual Domain Adaptation is a problem of immense importance in computer vision. Previous approaches showcase the inability of even deep neural networks to learn informative representations across domain shift. This problem is more severe for tasks where acquiring hand labeled data is extremely hard and tedious. In this work, we focus on adapting the representations learned by segmentation networks across synthetic and real domains. Contrary to previous approaches that use a simple adversarial objective or superpixel information to aid the process, we propose an approach based on Generative Adversarial Networks (GANs) that brings the embeddings closer in the learned feature space. To showcase the generality and scalability of our approach, we show that we can achieve state of the art results on two challenging scenarios of synthetic to real domain adaptation. Additional exploratory experiments show that our approach: (1) generalizes to unseen domains and (2) results in improved alignment of source and target distributions.

In the NIPS 2017 Learning to Run challenge, participants were tasked with building a controller for a musculoskeletal model to make it run as fast as possible through an obstacle course. Top participants were invited to describe their algorithms. In this work, we present eight solutions that used deep reinforcement learning approaches, based on algorithms such as Deep Deterministic Policy Gradient, Proximal Policy Optimization, and Trust Region Policy Optimization. Many solutions use similar relaxations and heuristics, such as reward shaping, frame skipping, discretization of the action space, symmetry, and policy blending. However, each of the eight teams implemented different modifications of the known algorithms.

Momentum is a simple and widely used trick which allows gradient-based optimizers to pick up speed in low curvature directions. Its performance depends crucially on a damping coefficient $\beta$. Large $\beta$ values can potentially deliver much larger speedups, but are prone to oscillations and instability; hence one typically resorts to small values such as 0.5 or 0.9. We propose Aggregated Momentum (AggMo), a variant of momentum which combines multiple velocity vectors with different $\beta$ parameters. AggMo is trivial to implement, but significantly dampens oscillations, enabling it to remain stable even for aggressive $\beta$ values such as 0.999. We reinterpret Nesterov's accelerated gradient descent as a special case of AggMo and provide theoretical convergence bounds for online convex optimization. Empirically, we find that AggMo is a suitable drop-in replacement for other momentum methods, and frequently delivers faster convergence.

We address a largely open problem of multilabel classification over graphs. Unlike traditional vector input, a graph has rich variable-size structures, that suggests complex relationships between labels and subgraphs. Uncovering these relations might hold the keys of classification performance and explainability. To this end, we design GAML (Graph Attentional Multi-Label learning), a graph neural network that models the relations present in the input graph, in the label set, and across graph-labels by leveraging the message passing algorithm and attention mechanism. Representation of labels and input nodes is refined iteratively through multiple steps, during which interesting subgraph-label patterns emerge. In addition, GAML is highly flexible by allowing explicit label dependencies to be incorporated easily. It also scales linearly with the number of labels and graph size thanks to our proposed hierarchical attention. These properties open a wide range of applications seen in the real world. We evaluate GAML on an extensive set of experiments with both graph inputs (for predicting drug-protein binding, and drug-cancer response), and classical unstructured inputs. The results are significantly better than well-known multilabel learning techniques.

Should radiologists be buying into the hype about artificial intelligence? HealthManagement spoke to Prof.Paul Chang about AI, deep learning and the advantages of intellectual arbitrage. How are artificial intelligence (AI) and deep learning shaping radiology? In healthcare in general, and radiology in particular, we tend to buy very early into the hype surrounding any new potentially disruptive technology, whether that's picture archiving and Communication systems (PACS), speech recognition or big data. But it takes us much longer to appropriately consume and actually influence'real-world' radiology.

Deep Learning (DL) is part of the field of Machine Learning (ML). DL works by approximating a solution to a problem using neural networks. One of the nice properties of about neural networks is that they find patterns in the data (features) by themselves. This is opposed to having to tell your algorithm what to look for, as in the olde times. However, often this means the model starts with a blank state (unless we are transfer learning).

Adobe and Nvidia have announced a partnership that will see both companies deliver new artificial intelligence (AI) and deep learning services for Adobe Creative. Making the announcement during the Adobe Summit keynote in Las Vegas on Wednesday, Adobe CEO and president Shantanu Narayen was joined by Nvidia founder and CEO Jensen Huang. Machine learning, task automation and robotics are already widely used in business. These and other AI technologies are about to multiply, and we look at how organizations can best take advantage of them. The CEOs said the partnership will see both companies work to optimise the Adobe Sensei AI and machine learning framework for Nvidia GPUs.