In his State of the Union address Tuesday, President Trump promised legislation to invest in "the cutting edge industries of the future." But the speech was characteristically backward-looking. Trump talked up gains in manufacturing jobs and oil and gas exports, but didn't once mention the word "technology," nor any other tech policy issue, such as privacy, broadband, or antitrust. Aides filled in the blanks. "President Trump's commitment to American leadership in artificial intelligence, 5G wireless, quantum science, and advanced manufacturing will ensure that these technologies serve to benefit the American people and that the American innovation ecosystem remains the envy of the world for generations to come," Michael Kratsios, deputy assistant to the president for technology policy, said in a statement.

As a competent marriage counselor might say, the foundation of any good relationship is communication--a principle the US military's research wing is taking to heart with a new program aimed at developing autonomous systems that are not just machines, but "trusted partners" with greater self-awareness. Autonomous technologies such as driverless cars cannot tell the humans they interact with much about how well they can perform in different conditions. A coach, for instance, might want to let a player with a sore arm sit out a game. Similarly, during bad weather, a passenger might choose a driverless car that performs better in the rain over another--if only the car could communicate its abilities better. The Defense Advanced Research Projects Agency (DARPA) seeks to develop that skill.

We study the problem of synthesizing strategies for a mobile sensor network to conduct surveillance in partnership with static alarm triggers. We formulate the problem as a multi-agent reactive synthesis problem with surveillance objectives specified as temporal logic formulas. In order to avoid the state space blow-up arising from a centralized strategy computation, we propose a method to decentralize the surveillance strategy synthesis by decomposing the multi-agent game into subgames that can be solved independently. We also decompose the global surveillance specification into local specifications for each sensor, and show that if the sensors satisfy their local surveillance specifications, then the sensor network as a whole will satisfy the global surveillance objective. Thus, our method is able to guarantee global surveillance properties in a mobile sensor network while synthesizing completely decentralized strategies with no need for coordination between the sensors. We also present a case study in which we demonstrate an application of decentralized surveillance strategy synthesis.

Abstract--Predicting flood for any location at times of extreme storms is a longstanding problem that has utmost importance in emergency management. Conventional methods that aim to predict water levels in streams use advanced hydrological models still lack of giving accurate forecasts everywhere. This study aims to explore artificial deep neural networks' performance on flood prediction. While providing models that can be used in forecasting stream stage, this paper presents a dataset that focuses on the connectivity of data points on river networks. It also shows that neural networks can be very helpful in time-series forecasting as in flood events, and support improving existing models through data assimilation. Hurricanes Harvey [1], Irma [2] and Maria, and other natural disasters caused more than $306 billion.

We ask whether the neural network interpretation methods can be fooled via adversarial model manipulation, which is defined as a model fine-tuning step that aims to radically alter the explanations without hurting the accuracy of the original model. By incorporating the interpretation results directly in the regularization term of the objective function for fine-tuning, we show that the state-of-the-art interpreters, e.g., LRP and Grad-CAM, can be easily fooled with our model manipulation. We propose two types of fooling, passive and active, and demonstrate such foolings generalize well to the entire validation set as well as transfer to other interpretation methods. Our results are validated by both visually showing the fooled explanations and reporting quantitative metrics that measure the deviations from the original explanations. We claim that the stability of neural network interpretation method with respect to our adversarial model manipulation is an important criterion to check for developing robust and reliable neural network interpretation method.

We consider adversarial examples in the black-box decision-based scenario. Here, an attacker has access to the final classification of a model, but not its parameters or softmax outputs. Most attacks for this scenario are based either on transferability, which is unreliable, or random sampling, which is often slow. Focusing on the latter, we propose to improve the efficiency of sampling-based attacks with prior beliefs about the target domain. We identify two such priors, image frequency and surrogate gradients, and discuss how to integrate them into a unified sampling procedure. We then formulate the Biased Boundary Attack, which achieves a drastic speedup over the original Boundary Attack. We demonstrate the effectiveness of our approach against an ImageNet classifier. We also showcase a targeted attack for the Google Cloud Vision API, where we craft convincing examples with just a few hundred queries. Finally, we demonstrate that our approach outperforms the state of the art when facing strong defenses: Our attack scored second place in the targeted attack track of the NeurIPS 2018 Adversarial Vision Challenge.

Video game sales were particularly cutthroat last holiday, and it appears EA's Battlefield V was one of the casualties. The publisher lowered the outlook for its revenue this quarter after revealing that its sales in the last quarter of 2018 "did not perform to our expectations." While it didn't explicitly blame BFV for the shortfall, that was undoubtedly the company's flagship game -- it otherwise relied on sports titles and the mobile-only Command & Conquer Rivals. It's not too hard to see why BFV would have struggled. EA delayed the release to November 20th, and left the the much-hyped battle royale mode out of the game until the spring.

Scientists are using the popular drawing game Pictionary to teach artificial intelligence common sense. AI researchers at the Allen Institute for Artificial Intelligence (AI2), a non-profit lab in Seattle, developed a version of the game called Iconary in order to teach its AllenAI artificial intelligence abstract concepts from pictures alone. Iconary was made public on 5 February in order to encourage people to play the game with AllenAI. By learning from humans, the researchers hope AllenAI will continue to develop common sense reasoning. "Iconary is one of the first times an AI system is paired in a collaborative game with a human player instead of antagonistically working against them," the Iconary website states.

The system, named Doctor Alzimov, after the Russian-born science fiction writer, can be installed on any computer and provides images with clearly marked findings for easy interpretation. It was developed by researchers at Peter the Great St. Petersburg Polytechnic University (SPbPU) in St. Petersburg, Russia along with radiologists from the St. Petersburg Oncological Center.

Artificial intelligence (AI) sizzled at the Radiological Society of North America's (RSNA) 2018 meeting as vendors promoted new, old and unconventional technologies as means to increase efficiency throughout imaging. New releases were unveiled in imaging modalities, notably magnetic resonance imaging (MRI) and computed tomography (CT). But excitement was muted compared to enthusiasm for AI, which has been high for two years. Interest in AI was stoked by multiple medical societies at their annual meetings in the weeks leading up to RSNA 2018. The Radiological Society of North America contributed further through its decision to set up a machine learning showcase in the back of the North exhibit hall. Enthusiasm for enterprise imaging (EI) increased with the vendor promotion of AI software that promised to streamline workflow. This "coattail effect" was seen in multimodality exhibits including those of GE, Siemens, Philips and Canon, which promoted the use of AI not just to improve scanners, but to boost the performance of picture archive and communication systems (PACS) and vendor neutral archives (VNA) as well.