A study by researchers at Bard College's Center for the Study of the Drone found that at least 95 countries currently own unmanned military drones. A report by researchers at Bard College's Center for the Study of the Drone estimated that 95 countries currently own unmanned military drones--up from 60 in 2010--and the infrastructure to support their operation also is expanding. Bard's Dan Gettinger said drones "are featuring more prominently in world affairs, as we've seen most recently in the Saudi drone attacks." The report found the U.S. monopoly on long-distance military drones is eroding, with at least 10 nations, including Azerbaijan and Nigeria, using drones to launch strikes. Fifteen countries have training academies for drone operators, according to the report.

Michael Waltz (R-FL) and Don Beyer (D-VA) produced a deepfake video for the U.S. House Science subcommittee to demonstrate the threat such disinformation presents. Michael Waltz (R-FL) and Don Beyer (D-VA) produced an artificial intelligence-doctored political video, or deepfake, for the U.S. House Science subcommittee to demonstrate the threat such disinformation presents. Lawmakers are worried of malefactors using deepfakes to disrupt and divide U.S. voters in the run-up to the 2020 election, and Waltz and Beyer are urging investment in deepfake-detection solutions, especially as production tools become increasingly affordable and accessible. State University of New York at Albany's Siwei Lyu, who helped craft the deepfake demo, said his software could generate deepfakes of a minute-long YouTube video in eight hours. Meanwhile, the University of California, Berkeley's Hany Farid cited the sluggish progress of technology platforms like Facebook and Google to address deepfakes.

ImageNet will remove 600,000 images of people stored on its database after an art project exposed racial bias in the program's artificial intelligence system. Created in 2009 by researchers at Princeton and Stanford, the online image database has been widely used by machine learning projects. The program has pulled more than 14 million images from across the web, which have been categorized by Amazon Mechanical Turk workers -- a crowdsourcing platform through which people can earn money performing small tasks for third parties. According to the results of an online project by AI researcher Kate Crawford and artist Trevor Paglen, prejudices in that labor pool appear to have biased the machine learning data. Training Humans -- an exhibition that opened last week at the Prada Foundation in Milan -- unveiled the duo's findings to the public, but part of their experiment also lives online at ImageNet Roulette, a website where users can upload their own photographs to see how the database might categorize them.

The U.S. Defense Department has made battlefield-ready artificial intelligence a priority in its planning, seeking a massive increase in related spending to counter China's rapid advances in the field. The department's Joint Artificial Intelligence Center requested $268 million under the draft federal budget for the fiscal year that began Oct. 1, roughly triple the figure from the previous year. "I am optimistic that 2020 will be a breakout year for the department when it comes to fielding AI-enabled capabilities," Lt. Gen. Jack Shanahan, the center's director, said in a recent press briefing. "For fiscal year '20, our biggest project will be what we are calling'AI for maneuver and fires,' with individual lines of effort or product lines oriented on warfighting operations," he told reporters. The U.S. military's AI development has focused on predictive maintenance for weapons systems, along with areas such as humanitarian missions and cybersecurity.

Researchers and policymakers have warned that if the autonomous systems we're increasingly integrating into our lives are not carefully designed, there could be serious repercussions. But there are also opportunities to ensure that human values stay at the center of new advances in artificial intelligence. At this critical moment, Northeastern is launching the Institute for Experiential Artificial Intelligence, a pioneering research hub that will place human skills and intelligence at the forefront of artificial intelligence development, from the earliest design steps to the final implementation. The university is allocating $50 million for the new institute, in which leading experts in the humanities, law, public policy, machine learning, health, security, and sustainability will collaborate to develop solutions to the world's challenges. "This new institute, the first of its kind, will focus on enabling artificial intelligence and humans to collaborate interactively around solving problems in health, security, and sustainability," Northeastern President Joseph E. Aoun wrote to the university's students, faculty, and staff.

In recent years, our fascination with the potential of AI has taken a more starry-eyed turn, as shown in the 2013 sci-fi drama "Her," where the main character falls in love with a virtual assistant. In reality, artificial intelligence (AI) technology is quickly permeating every aspect of our lives. From Amazon's voice-activated Alexa to writing technology that helps managers craft job postings, AI is in our hearts, homes and workplaces. And it's only going to become a bigger part of our lives: Experts call the rise of AI the driving force behind the fourth industrial revolution. On a recent afternoon at the NVIDIA robotics research lab in Seattle's University District, researchers use a simulated kitchen to test robots' ability to perform simple tasks such as grabbing objects.

Information theoretic criteria (ITC) have been widely adopted in engineering and statistics for selecting, among an ordered set of candidate models, the one that better fits the observed sample data. The selected model minimizes a penalized likelihood metric, where the penalty is determined by the criterion adopted. While rules for choosing a penalty that guarantees a consistent estimate of the model order are known, theoretical tools for its design with finite samples have never been provided in a general setting. In this paper, we study model order selection for finite samples under a design perspective, focusing on the generalized information criterion (GIC), which embraces the most common ITC. The theory is general, and as case studies we consider: a) the problem of estimating the number of signals embedded in additive white Gaussian noise (AWGN) by using multiple sensors; b) model selection for the general linear model (GLM), which includes e.g. the problem of estimating the number of sinusoids in AWGN. The analysis reveals a trade-off between the probabilities of overestimating and underestimating the order of the model. We then propose to design the GIC penalty to minimize underestimation while keeping the overestimation probability below a specified level. For the considered problems, this method leads to analytical derivation of the optimal penalty for a given sample size. A performance comparison between the penalty optimized GIC and common AIC and BIC is provided, demonstrating the effectiveness of the proposed design strategy.

The regression discontinuity (RD) design is one of the most popular quasi-experimental methods for applied causal inference. In practice, the method is quite sensitive to the assumption that individuals cannot control their value of a "running variable" that determines treatment status precisely. If individuals are able to precisely manipulate their scores, then point identification is lost. We propose a procedure for obtaining partial identification bounds in the case of a discrete running variable where manipulation is present. Our method relies on two stages: first, we derive the distribution of non-manipulators under several assumptions about the data. Second, we obtain bounds on the causal effect via a sequential convex programming approach. We also propose methods for tightening the partial identification bounds using an auxiliary covariate, and derive confidence intervals via the bootstrap. We demonstrate the utility of our method on a simulated dataset.

Motivated by safety-critical classification problems, we investigate adversarial attacks against cost-sensitive classifiers. We use current state-of-the-art adversarially-resistant neural network classifiers [1] as the underlying models. Cost-sensitive predictions are then achieved via a final processing step in the feed-forward evaluation of the network. We evaluate the effectiveness of cost-sensitive classifiers against a variety of attacks and we introduce a new cost-sensitive attack which performs better than targeted attacks in some cases. We also explored the measures a defender can take in order to limit their vulnerability to these attacks. This attacker/defender scenario is naturally framed as a two-player zero-sum finite game which we analyze using game theory.

Access to a large amount of high quality data is possibly the most important factor for success in advancing medicine with machine learning and data science. However, valuable healthcare data are usually distributed across isolated silos, and there are complex operational and regulatory concerns. Data on patient populations are often horizontally separated,each other across different practices and health systems. In addition, individual patient data are often vertically separated, by data type, across her sites of care, service, and testing. We train a confederated learning model in a manner to stratify elderly patients by their risk of a fall in the next two years, using diagnoses, medication claims data and clinical lab test records of patients.