Can we automatically design a Convolutional Network (ConvNet) with the highest image classification accuracy under the latency constraint of a mobile device? Neural Architecture Search (NAS) for ConvNet design is a challenging problem due to the combinatorially large design space and search time (at least 200 GPU-hours). To alleviate this complexity, we propose Single-Path NAS, a novel differentiable NAS method for designing device-efficient ConvNets in less than 4 hours. 1. Novel NAS formulation: our method introduces a single-path, over-parameterized ConvNet to encode all architectural decisions with shared convolutional kernel parameters. 2. NAS efficiency: Our method decreases the NAS search cost down to 8 epochs (30 TPU-hours), i.e., up to 5,000x faster compared to prior work. 3. On-device image classification: Single-Path NAS achieves 74.96% top-1 accuracy on ImageNet with 79ms inference latency on a Pixel 1 phone, which is state-of-the-art accuracy compared to NAS methods with similar latency (<80ms).

In reinforcement learning algorithms, it is a common practice to account for only a single view of the environment to make the desired decisions; however, utilizing multiple views of the environment can help to promote the learning of complicated policies. Since the views may frequently suffer from partial observability, their provided observation can have different levels of importance. In this paper, we present a novel attention-based deep reinforcement learning method in a multi-view environment in which each view can provide various representative information about the environment. Specifically, our method learns a policy to dynamically attend to views of the environment based on their importance in the decision-making process. We evaluate the performance of our method on TORCS racing car simulator and three other complex 3D environments with obstacles.

Reinforcement learning (RL) methods learn optimal decisions in the presence of a stationary environment. However, the stationary assumption on the environment is very restrictive. In many real world problems like traffic signal control, robotic applications, one often encounters situations with non-stationary environments and in these scenarios, RL methods yield sub-optimal decisions. In this paper, we thus consider the problem of developing RL methods that obtain optimal decisions in a non-stationary environment. The goal of this problem is to maximize the long-term discounted reward achieved when the underlying model of the environment changes over time. To achieve this, we first adapt a change point algorithm to detect change in the statistics of the environment and then develop an RL algorithm that maximizes the long-run reward accrued. We illustrate that our change point method detects change in the model of the environment effectively and thus facilitates the RL algorithm in maximizing the long-run reward. We further validate the effectiveness of the proposed solution on non-stationary random Markov decision processes, a sensor energy management problem and a traffic signal control problem.

Value iteration is a fixed point iteration technique utilized to obtain the optimal value function and policy in a discounted reward Markov Decision Process (MDP). Here, a contraction operator is constructed and applied repeatedly to arrive at the optimal solution. Value iteration is a first order method and therefore it may take a large number of iterations to converge to the optimal solution. In this work, we propose a novel second order value iteration procedure based on the Newton-Raphson method. We first construct a modified contraction operator and then apply Newton-Raphson method to arrive at our algorithm. We prove the global convergence of our algorithm to the optimal solution and show the second order convergence. Through experiments, we demonstrate the effectiveness of our proposed approach.

In this paper, we apply a new promising tool for pattern classification, namely, the Tsetlin Machine (TM), to the field of disease forecasting. The TM is interpretable because it is based on manipulating expressions in propositional logic, leveraging a large team of Tsetlin Automata (TA). Apart from being interpretable, this approach is attractive due to its low computational cost and its capacity to handle noise. To attack the problem of forecasting, we introduce a preprocessing method that extends the TM so that it can handle continuous input. Briefly stated, we convert continuous input into a binary representation based on thresholding. The resulting extended TM is evaluated and analyzed using an artificial dataset. The TM is further applied to forecast dengue outbreaks of all the seventeen regions in Philippines using the spatio-temporal properties of the data. Experimental results show that dengue outbreak forecasts made by the TM are more accurate than those obtained by a Support Vector Machine (SVM), Decision Trees (DTs), and several multi-layered Artificial Neural Networks (ANNs), both in terms of forecasting precision and F1-score.

Partial domain adaptation aims to transfer knowledge from a label-rich source domain to a label-scarce target domain which relaxes the fully shared label space assumption across different domains. In this more general and practical scenario, a major challenge is how to select source instances in the shared classes across different domains for positive transfer. To address this issue, we propose a Domain Adversarial Reinforcement Learning (DARL) framework to automatically select source instances in the shared classes for circumventing negative transfer as well as to simultaneously learn transferable features between domains by reducing the domain shift. Specifically, in this framework, we employ deep Q-learning to learn policies for an agent to make selection decisions by approximating the action-value function. Moreover, domain adversarial learning is introduced to learn domain-invariant features for the selected source instances by the agent and the target instances, and also to determine rewards for the agent based on how relevant the selected source instances are to the target domain. Experiments on several benchmark datasets demonstrate that the superior performance of our DARL method over existing state of the arts for partial domain adaptation.

Millions of people's private photos have been leveraged by the cloud photo service, Ever, to develop and sell facial recognition software without their consent says an exclusive report by NBC News. According to the report, Ever, which started in 2013 as a cloud-based app for storing and sharing photos, has recently started to pivot into a burgeoning field of facial recognition technology through its new arm, Ever AI. In order to train its software, which according to the company's web page, is capable of delivering'surveillance & monitoring, physical access control, and digital authentication,' it used the personal photos from its millions of its users without informing them first. According to the privacy policy and a statement from CEO of Ever, Doug Aley, the company does not distribute users' photos to third parties, but does use them to instruct its algorithm. Specifically, it leverages a facial recognition feature built into the Ever service which allows users to group photos of the same people by scanning their face.

Toyota Motor Corp. and Panasonic Corp. said Thursday they will integrate their housing businesses in an expansion of an existing tie-up, as they seek to collaborate on "town development" for next-generation lifestyles where homes and vehicles are connected to the internet. The companies plan to set up a joint venture on Jan. 7, 2020. Toyota will focus on new mobility services using self-driving technology, while Panasonic brings strengths in developing smart homes equipped with appliances supported by internet and other digital technologies. The venture, Prime Life Technologies Corp., will bring housing units of both companies under its wing, including Toyota Housing Corp., Misawa Homes Co. and Panasonic Homes Co. The move by the leading carmaker and electronics manufacturer comes amid shrinking demand in the domestic housing market due to Japan's declining population.

American University of Beirut is developing a tool that farmers in the Middle East and Africa can use to irrigate fields at the optimum times to save water. At Colegio Mayor de Nuestra Señora del Rosario, a university in Colombia, researchers will use satellite images to detect illegal mines that are polluting community drinking water. Crisis Text Line, a nonprofit that connects people experiencing a crisis with volunteer counselors by text message, uses AI to evaluate messages and move the people who are in most danger to the front of the line. In Australia, a public health service called Eastern Health will use AI to comb through clinical records from ambulances and find patterns in suicide attempts–and ways to intervene earlier. Full Fact, an independent fact-checking organization in the U.K., is using AI to help human fact-checkers more quickly assess claims made by politicians and the media.

Solutions that can generate accurate estimates of counts are in demand, whether it is for tallying the number of people in a video frame, counting the number of animals of an endangered species, estimating the number of objects or shapes in a picture, or for a variety of similar industry applications. Traditional crowd counting methods and models that use detection or regression-based approaches have been plagued by challenges such as occlusion, non-uniform distribution, perspective distortion, camera angles and background clutter. They are not robust and often fail with even simple changes to the planned scenarios. Deep learning based crowd counting solutions offer an excellent recourse to such problems. Cascaded CNN's use density-based estimations to preserve the spatial information and can localize the count, in addition to estimating the overall tally.