China is on its way to becoming the first global superpower for Artificial Intelligence. The People's Republic of China has the most ambitious AI strategy of all nations and provides the most resources worldwide for its implementation. China combines a gigantic amount of data with talent, companies, research and capital to build the world's leading AI ecosystem. In 2017, the State Council of the People's Republic of China (also known as the Central People's Government) published the Artificial Intelligence Development Plan (here you can find the original document in English). This strategy is part of the even bigger national "Made in China 2025″ plan and will also be linked to the new (digital) Silk Road.

Mysterious drone swarms have been seen flying in Colorado, Nebraska and Wyoming at night since December, sometimes over locations believed to house nuclear missile silos. A federal task force has been formed to investigate the drones' origin and purpose. The Phillips County Sheriff's Office in Colorado reported the first drones on 20 December. There have been hundreds of sightings since, some of groups of drones flying in grid patterns. Some observers assumed the drones were part of a military exercise, but the US Air Force has denied involvement.

WASHINGTON – The United States on Wednesday resumed joint military operations with Iraq that had been put on pause after the U.S. drone strike that killed a top Iranian general in Baghdad, The New York Times reported. Two U.S. military officials quoted by the paper said the Pentagon wanted to resume these operations in order to pick up the fight against the Islamic State group. Washington began the pause on January 5 two days after the strike that killed Iranian Gen. Qassem Soleimani at the Baghdad airport. The same day of the suspension furious Iraqi lawmakers voted to expel the more than 5,000 U.S. troops that are in Iraq. It was not immediately clear if anyone in the Iraqi government had approved the resumption of the joint military operations, the Times reported. The Pentagon said it had no information to provide concerning a resumption when contacted by AFP.

The U.S. military is spending more than $4.5 million to develop facial recognition technology that reads the pattern of heat being emitted by faces in order to identify specific people. The technology would work in the dark and across long distances, according to contracts posted on a federal spending database. Facial recognition is already employed by the military, which uses the technology to identify individuals on the battlefield. But existing facial recognition technology typically relies on images generated by standard cameras, such as those found in iPhone or CCTV networks. Now, the military wants to develop a facial recognition system that analyzes infrared images to identify individuals.

This idea of using disruptive camouflage was extensively undertaken during World War I and also somewhat during World War II. We tend to think of navy ships as always being painted a rather dull monotone grey color. This would seem to be a wise choice. At sea, the navy ships would tend to blend into the background of a grayish sky and a blue sea. Presumably, whales and dolphins use a similar colorization to try to blend into their surroundings.

As the U.S. Army increasingly uses facial and object recognition to train artificial intelligent systems to identify threats, the need to protect its systems from cyberattacks becomes essential. An Army project conducted by researchers at Duke University and led by electrical and computer engineering faculty members Dr. Helen Li and Dr. Yiran Chen, made significant progress toward mitigating these types of attacks. Two members of the Duke team, Yukun Yang and Ximing Qiao, recently took first prize in the Defense category of the CSAW '19 HackML competition (see Related Links below). "Object recognition is a key component of future intelligent systems, and the Army must safeguard these systems from cyberattacks," said MaryAnne Fields, program manager for intelligent systems at the Army Research Office. "This work will lay the foundations for recognizing and mitigating backdoor attacks in which the data used to train the object recognition system is subtly altered to give incorrect answers. Safeguarding object recognition systems will ensure that future Soldiers will have confidence in the intelligent systems they use."

When we get to a point where literally just about everything can be done more cheaply and more efficiently by robots, the elite won't have any use for the rest of us at all. For most of human history, the wealthy have needed the poor to do the work that is necessary to run their businesses and make them even wealthier. In this day and age we like to call ourselves "employees", but in reality we are their servants. Some of us may be more well paid than others, but the vast majority of us are expending our best years serving their enterprises so that we can pay the bills. Unfortunately, that paradigm is rapidly changing, and many of the jobs that humans are doing today will be done by robots in the not too distant future.

Deep learning (DL) is transforming whole industries as complicated decision-making processes are being automated by Deep Neural Networks (DNNs) trained on real-world data. Driven in part by a rapidly-expanding literature on DNN approximation theory showing that DNNs can approximate a rich variety of functions, these tools are increasingly being considered for problems in scientific computing. Yet, unlike more traditional algorithms in this field, relatively little is known about DNNs from the principles of numerical analysis, namely, stability, accuracy, computational efficiency and sample complexity. In this paper we introduce a computational framework for examining DNNs in practice, and use it to study their empirical performance with regard to these issues. We examine the performance of DNNs of different widths and depths on a variety of test functions in various dimensions, including smooth and piecewise smooth functions. We also compare DL against best-in-class methods for smooth function approximation based on compressed sensing. Our main conclusion is that there is a crucial gap between the approximation theory of DNNs and their practical performance, with trained DNNs performing relatively poorly on functions for which there are strong approximation results (e.g. smooth functions), yet performing well in comparison to best-in-class methods for other functions. Finally, we present a novel practical existence theorem, which asserts the existence of a DNN architecture and training procedure which offers the same performance as current best-in-class schemes. This result indicates the potential for practical DNN approximation, and the need for future research into practical architecture design and training strategies.

Adversarial attacks on deep neural networks (DNNs) have been found for several years. However, the existing adversarial attacks have high success rates only when the information of the attacked DNN is well-known or could be estimated by structure similarity or massive queries. In this paper, we propose an \emph{Attack on Attention} (AoA), a semantic feature commonly shared by DNNs. The transferability of AoA is quite high. With no more than 10 queries of the decision only, AoA can achieve almost 100\% success rate when attacking on many popular DNNs. Even without query, AoA could keep a surprisingly high attack performance. We apply AoA to generate 96020 adversarial samples from ImageNet to defeat many neural networks, and thus name the dataset as \emph{DAmageNet}. 20 well-trained DNNs are tested on DAmageNet. Without adversarial training, most of the tested DNNs have an error rate over 90\%. DAmageNet is the first universal adversarial dataset and it could serve as a benchmark for robustness testing and adversarial training.

The examples are considered to be of the form ( x i,y i), where x i is the feature vector of the i-th example and y i { 1, 1} is its class label. Extension to the multiclass case is often handled by breaking down the problem into multiple binary ones, so our analysis and its main results can carry over to the multiclass case. We consider the online setting where examples are presented to the learner in M minibatches 2 of size b. On the n -th iteration the learner performs the following steps: 1. Receive new examples x i, x i minibatch n 2. Predict the label ˆ y i and/or the probability estimate ˆ p(y i 1 x i), i minibatch n 3. Get true labels y i f ( x i), x i minibatch n, where f is the labelling function 4. Update learner parameters accordingly The steps above are intentionally left general enough to describe all learning components encountered in the paper. Our goal is to study the quality of the probability estimates generated by online boosting ensembles and strategies for improving it. Online boosting ensembles consist of multiple base learners, themselves also trained in an online fashion and -as we will seethe techniques used for improving the probability estimates (both the calibrator and the reward models of the bandits) are also learners trained in an online fashion. All follow the same general approach defined above: they maintain a model with a fixed number of parameters (i.e.