We consider the problem of fairly allocating a set of indivisible goods among n agents. Various fairness notions have been proposed within the rapidly growing field of fair division, but the Nash social welfare (NSW) serves as a focal point. In part, this follows from the ‘unreasonable’ fairness guarantees provided, in the sense that a max NSW allocation meets multiple other fairness metrics simultaneously, all while satisfying a standard economic concept of efficiency, Pareto optimality. However, existing approximation algorithms fail to satisfy all of the remarkable fairness guarantees offered by a max NSW allocation, instead targeting only the specific NSW objective. We address this issue by presenting a 2 max NSW, Prop-1, 1/(2n) MMS, and Pareto optimal allocation in strongly polynomial time. Our techniques are based on a market interpretation of a fractional max NSW allocation. We present novel definitions of fairness concepts in terms of market prices, and design a new scheme to round a market equilibrium into an integral allocation in a way that provides most of the fairness properties of an integral max NSW allocation.

The promise of reinforcement learning is to solve complex sequential decision problems by specifying a high-level reward function only. However, RL algorithms struggle when, as is often the case, simple and intuitive rewards provide sparse and deceptive feedback. Avoiding these pitfalls requires thoroughly exploring the environment, but despite substantial investments by the community, creating algorithms that can do so remains one of the central challenges of the field. We hypothesize that the main impediment to effective exploration originates from algorithms forgetting how to reach previously visited states ("detachment") and from failing to first return to a state before exploring from it ("derailment"). We introduce Go-Explore, a family of algorithms that addresses these two challenges directly through the simple principles of explicitly remembering promising states and first returning to such states before exploring. Go-Explore solves all heretofore unsolved Atari games (those for which algorithms could not previously outperform humans when evaluated following current community standards) and surpasses the state of the art on all hard-exploration games, with orders of magnitude improvements on the grand challenges Montezuma's Revenge and Pitfall. We also demonstrate the practical potential of Go-Explore on a challenging and extremely sparse-reward robotics task. Additionally, we show that adding a goal-conditioned policy can further improve Go-Explore's exploration efficiency and enable it to handle stochasticity throughout training. The striking contrast between the substantial performance gains from Go-Explore and the simplicity of its mechanisms suggests that remembering promising states, returning to them, and exploring from them is a powerful and general approach to exploration, an insight that may prove critical to the creation of truly intelligent learning agents.

We consider a system consisting of a single transmitter/receiver pair and $N$ channels over which they may communicate. Packets randomly arrive to the transmitter's queue and wait to be successfully sent to the receiver. The transmitter may attempt a frame transmission on one channel at a time, where each frame includes a packet if one is in the queue. For each channel, an attempted transmission is successful with an unknown probability. The transmitter's objective is to quickly identify the best channel to minimize the number of packets in the queue over $T$ time slots. To analyze system performance, we introduce queue length regret, which is the expected difference between the total queue length of a learning policy and a controller that knows the rates, a priori. One approach to designing a transmission policy would be to apply algorithms from the literature that solve the closely-related stochastic multi-armed bandit problem. These policies would focus on maximizing the number of successful frame transmissions over time. However, we show that these methods have $\Omega(\log{T})$ queue length regret. On the other hand, we show that there exists a set of queue-length based policies that can obtain order optimal $O(1)$ queue length regret. We use our theoretical analysis to devise heuristic methods that are shown to perform well in simulation.

Robot in Singapore is cracking down on people's social distancing Singapore has employed a four-legged robot named SPOT to patrol a popular park and remind people to maintain social distancing.

Artificial intelligence caught Shashank Ojha's imagination while he was a student at Thomas Jefferson High School for Science and Technology in Alexandria, Virginia. He took the few AI courses the school offered and soon set his sights on attending Carnegie Mellon University. "I knew that CMU was the place to be for AI," he explained. His plan when he entered CMU in 2016 was to pursue a bachelor's degree in computer science, with minors in machine learning and robotics. What he hadn't counted on was the School of Computer Science's 2018 decision to launch the nation's first undergraduate AI degree program.

A smartphone app may be able to diagnose coronavirus by using the microphone in the device to measure changes in breathing sounds, the developers claim. Researchers from the University of Pennsylvania say, if successful, their work would provide a'low cost solution' to testing an entire population quickly. It is still at the planning stage so it could be some time before it is available in app stores - the team say they first have to design new'acoustic waveforms'. The technology will use artificial intelligence to examine the sound coming from human airways and determine if it matches those from COVID-19 patients. When developed the app will use existing hardware and computing power of normal smartphones to provide'non-invasive at-home testing', developers claim.

NASA has created a new'mini' Mars rover that will be able to climb hills even if they're covered in sand and gravel - perfect for future exploration missions. Driven by the knowledge that the rolling hills of Mars are a long way from the nearest tow truck, Georgia Tech researchers created a more resilient'wiggling' vehicle. Unlike the NASA Curiosity rover currently on Mars, or NASA Perseverance launching this summer, this'mini' design can walk, paddle and wheelspin its way up a hill. The team created the mini vehicle based on the NASA Resource Prospector 15 (RP15) rover that can wiggle its wheels to enhance and test the design. Professor Dan Goldman, from the School of Physics at the Georgia Institute of Technology, said the rover was built using 3D printed parts.

According to experts, there is very little chance a vaccine for COVID-19 will be perfected and ready for use this year. This eagerly sought goal might take place by 2021 at the earliest -- but only if things proceed smoothly with the 60 vaccine trials currently taking place. The latest expert source to attest to this impossibility used its experience in quantitative financial investment to analyze the progress being made by the 60 vaccine candidates. Boston-based PanAgora Asset Management analyzed vast quantities of medical research data to calculate which of the 60 will succeed in producing a successful vaccine within the year. The quick answer is "zero."

Eurovision 2020 was unsurprisingly cancelled due to the pandemic, but AI has stepped in to fill its glittery shoes. Dutch broadcaster VPRO has just wrapped up a Eurovision-inspired AI Song Contest, with 13 teams from Europe and Australia training algorithms to become budding pop stars while experts judge their output. As BBC and Bloomberg point out, the results are a mix of surprisingly well-done and frighteningly dystopic tunes... a bit like the real thing, really. The winning entry came from Australian team Uncanny Valley, whose song "Beautiful the World" was built by an AI trained on a mix of Eurovision hits and local animals affected by wildfires, including koalas, kookaburras and Tasmanian devils. It has the same catchy dance pop riffs you'd expect to get the full douze points from a Eurovision vote, just with nonsensical lyrics.

Jim Sterne is widely considered the godfather of digital marketing analytics and is one of our heroes in the world of marketing and data. Jim was along every step of the way as tech evolved in Silicon Valley--from selling companies their first-ever computers, to imagining what the World Wide Web could mean for business in the 90s, to his current focus on how Artificial Intelligence and Machine Learning is going to be the next ubiquitous element of running a business, in particular for the fields of data analysis and marketing. Jim has written a whole shelf's worth of books, several of which he found himself writing because no one else had. His latest book is one of these, entitled, "Artificial Intelligence for Marketing: Practical Applications." He also founded the Marketing Analytics Summit and is gearing up for this year's conference on June 17-20--the event where digital analysts, marketing technologists, and growth architects use marketing data to create impact.