We present three results on the complexity of Minimax Approval Voting. First, we study Minimax Approval Voting parameterized by the Hamming distance d from the solution to the votes. We show Minimax Approval Voting admits no algorithm running in time O*(2o(d log d)), unless the Exponential Time Hypothesis (ETH) fails. This means that the O*(d2d) algorithm of Misra, Nabeel and Singh is essentially optimal. Motivated by this, we then show a parameterized approximation scheme, running in time O*((3/ε)2d), which is essentially tight assuming ETH. Finally, we get a new polynomial-time randomized approximation scheme for Minimax Approval Voting, which runs in time nO(1/ε2⋅log(1/ε))⋅poly(m), where n is a number of voters and m is a number of alternatives. It almost matches the running time of the fastest known PTAS for Closest String due to Ma and Sun.

Transfer learning, which allows a source task to affect the inductive bias of the target task, is widely used in computer vision. The typical way of conducting transfer learning with deep neural networks is to fine-tune a model pre-trained on the source task using data from the target task. In this paper, we propose an adaptive fine-tuning approach, called SpotTune, which finds the optimal fine-tuning strategy per instance for the target data. In SpotTune, given an image from the target task, a policy network is used to make routing decisions on whether to pass the image through the fine-tuned layers or the pre-trained layers. We conduct extensive experiments to demonstrate the effectiveness of the proposed approach. Our method outperforms the traditional fine-tuning approach on 12 out of 14 standard datasets.We also compare SpotTune with other state-of-the-art fine-tuning strategies, showing superior performance. On the Visual Decathlon datasets, our method achieves the highest score across the board without bells and whistles.

Grey-box fuzzers such as American Fuzzy Lop (AFL) are popular tools for finding bugs and potential vulnerabilities in programs. While these fuzzers have been able to find vulnerabilities in many widely used programs, they are not efficient; of the millions of inputs executed by AFL in a typical fuzzing run, only a handful discover unseen behavior or trigger a crash. The remaining inputs are redundant, exhibiting behavior that has already been observed. Here, we present an approach to increase the efficiency of fuzzers like AFL by applying machine learning to directly model how programs behave. We learn a forward prediction model that maps program inputs to execution traces, training on the thousands of inputs collected during standard fuzzing. This learned model guides exploration by focusing on fuzzing inputs on which our model is the most uncertain (measured via the entropy of the predicted execution trace distribution). By focusing on executing inputs our learned model is unsure about, and ignoring any input whose behavior our model is certain about, we show that we can significantly limit wasteful execution. Through testing our approach on a set of binaries released as part of the DARPA Cyber Grand Challenge, we show that our approach is able to find a set of inputs that result in more code coverage and discovered crashes than baseline fuzzers with significantly fewer executions.

Amid the country's growing substance abuse crisis, last year the FDA cleared reSET, a mobile app that tracks substance use, cravings, and social triggers to treat dependency on alcohol, cocaine, and cannabis. The FDA's clearance makes reSET one of the first prescription "digital therapeutics"--an emerging class of evidence-based interventions that are predominantly driven by software rather than drugs. Andy Coravos is is the CEO of Elektra Labs and a member of the Harvard-MIT Center for Regulatory Science. Earlier this year, digital medicine company Akili Interactive announced that its video game for children with ADHD demonstrated a statistically significant improvement in a randomized, controlled clinical trial. That milestone paves the way for what could be the first prescription video game.

A partnership between area public safety agencies and a startup technology firm has led to increased safety on a stretch of Las Vegas' busiest roadway. The year-long crash prevention pilot program between Waycare, a provider of artificial intelligence-based mobility solutions for smart cities, the Regional Transportation Commission of Southern Nevada, the Nevada Department of Transportation and the Nevada Highway Patrol resulted in a 17 percent reduction of crashes along a stretch of Interstate 15 northbound, the RTC announced. A portion of I-15 northbound between Charleston Boulevard and Russell Road was included in the program's results. As resources become available the program will be utilized in other areas of the city, including a stretch of U.S. Highway 95 between I-15 and the Rainbow Boulevard curve. The results came without any additional resources from state or local agencies, as the program utilizes in-vehicle information, existing cameras, sensors and other traffic-related data to develop predictive analytics that Waycare claims reduces traffic fatalities and crash related congestion.

Drumroll, please... after five years of deliberation over more than 60 candidate sites, NASA has finally decided where the Mars 2020 rover will land. An artist's conception shows the descent of NASA's Curiosity rover to the Martian surface. Humans won't be on it, but now we know where NASA's next mission to search for signs of life on Mars will land. After searching and debating the possibilities for five years, NASA chose the ancient Jezero Crater lakebed as the landing site for its upcoming Mars 2020 robot rover mission, the agency announced Monday. "The landing site in Jezero Crater offers geologically rich terrain, with landforms reaching as far back as 3.6 billion years old, that could potentially answer important questions in planetary evolution and astrobiology," NASA's Thomas Zurbuchen said in a statement.

Should we be scared of artificial intelligence (AI)? Some notable individuals such as legendary physicist Stephen Hawking and Tesla and SpaceX leader and innovator Elon Musk suggest AI could potentially be very dangerous; Musk at one point was comparing AI to the dangers of the dictator of North Korea. Microsoft co-founder Bill Gates also believes there's reason to be cautious, but that the good can outweigh the bad if managed properly. Since recent developments have made super-intelligent machines possible much sooner than initially thought, the time is now to determine what dangers artificial intelligence poses. What is applied and generalized artificial intelligence?

China's Ministry of Industry and Information Technology (MIIT) has identified 17 key areas as priority for artificial intelligence (AI) development through inviting private enterprises and research institutions to lead key projects and set up near-term targets for 2020, as the country is determined to compete with the US in the global AI races with an aim to lead the competition by 2030. What is AI? Everything you need to know about Artificial Intelligence The 17 key areas for AI development in China include intelligent vehicles, intelligent service robots, intelligent drones, neural network chips, and intelligent manufacturing among others, the official ministry said last week according to a Sina news report. The move aims to "select a group of enterprises and scientific research institutions that have mastered the key AI technologies, have strong innovation capabilities, and have great development potential", to break the bottleneck in China's AI development, cultivate a force of enterprises in innovation and development, accelerate deep integration between China's AI industry and economy, and promote innovation and development, according to the notice. By 2020, China are hoping to make breakthroughs in key technologies such as smart chips for autonomous driving, vehicle intelligent algorithms, and vehicle communications. Intelligent service robots should be able to achieve environment perception, natural interaction, independent learning and human-machine collaboration, with China hoping that they can be mass produced for home and public service purposes, according to the notice.

We propose an exploration method that incorporates look-ahead search over basic learnt skills and their dynamics, and use it for reinforcement learning (RL) of manipulation policies . Our skills are multi-goal policies learned in isolation in simpler environments using existing multigoal RL formulations, analogous to options or macroactions. Coarse skill dynamics, i.e., the state transition caused by a (complete) skill execution, are learnt and are unrolled forward during lookahead search. Policy search benefits from temporal abstraction during exploration, though itself operates over low-level primitive actions, and thus the resulting policies does not suffer from suboptimality and inflexibility caused by coarse skill chaining. We show that the proposed exploration strategy results in effective learning of complex manipulation policies faster than current state-of-the-art RL methods, and converges to better policies than methods that use options or parametrized skills as building blocks of the policy itself, as opposed to guiding exploration. We show that the proposed exploration strategy results in effective learning of complex manipulation policies faster than current state-of-the-art RL methods, and converges to better policies than methods that use options or parameterized skills as building blocks of the policy itself, as opposed to guiding exploration.

Neural networks are vulnerable to adversarial examples, malicious inputs crafted to fool trained models. Adversarial examples often exhibit black-box transfer, meaning that adversarial examples for one model can fool another model. However, adversarial examples may be overfit to exploit the particular architecture and feature representation of a source model, resulting in sub-optimal black-box transfer attacks to other target models. This leads us to introduce the Intermediate Level Attack (ILA), which attempts to fine-tune an existing adversarial example for greater black-box transferability by increasing its perturbation on a pre-specified layer of the source model. We show that our method can effectively achieve this goal and that we can decide a nearly-optimal layer of the source model to perturb without any knowledge of the target models.