Self-similarity refers to the image prior widely used in image restoration algorithms that small but similar patterns tend to occur at different locations and scales. However, recent advanced deep convolutional neural network based methods for image restoration do not take full advantage of self-similarities by relying on self-attention neural modules that only process information at the same scale. To solve this problem, we present a novel Pyramid Attention module for image restoration, which captures long-range feature correspondences from a multi-scale feature pyramid. Inspired by the fact that corruptions, such as noise or compression artifacts, drop drastically at coarser image scales, our attention module is designed to be able to borrow clean signals from their "clean" correspondences at the coarser levels. The proposed pyramid attention module is a generic building block that can be flexibly integrated into various neural architectures. Its effectiveness is validated through extensive experiments on multiple image restoration tasks: image denoising, demosaicing, compression artifact reduction, and super resolution. Without any bells and whistles, our PANet (pyramid attention module with simple network backbones) can produce state-of-the-art results with superior accuracy and visual quality. Our code will be available at https://github.com/SHI-Labs/Pyramid-Attention-Networks

We consider representation learning (hypothesis class $\mathcal{H} = \mathcal{F}\circ\mathcal{G}$) where training and test distributions can be different. Recent studies provide hints and failure examples for domain invariant representation learning, a common approach for this problem, but the explanations provided are somewhat different and do not provide a unified picture. In this paper, we provide new decompositions of risk which give finer-grained explanations and clarify potential generalization issues. For Single-Source Domain Adaptation, we give an exact decomposition (an equality) of the target risk, via a natural hybrid argument, as sum of three factors: (1) source risk, (2) representation conditional label divergence, and (3) representation covariate shift. We derive a similar decomposition for the Multi-Source case. These decompositions reveal factors (2) and (3) as the precise reasons for failure to generalize. For example, we demonstrate that domain adversarial neural networks (DANN) attempt to regularize for (3) but miss (2), while a recent technique Invariant Risk Minimization (IRM) attempts to account for (2) but does not consider (3). We also verify our observations experimentally.

We examine the applicability of modern neural network architectures to the midterm prediction of earthquakes. Our data-based classification model aims to predict if an earthquake with the magnitude above a threshold takes place at a given area of size $10 \times 10$ kilometers in $10$-$60$ days from a given moment. Our deep neural network model has a recurrent part (LSTM) that accounts for time dependencies between earthquakes and a convolutional part that accounts for spatial dependencies. Obtained results show that neural networks-based models beat baseline feature-based models that also account for spatio-temporal dependencies between different earthquakes. For historical data on Japan earthquakes our model predicts occurrence of an earthquake in $10$ to $60$ days from a given moment with magnitude $M_c > 5$ with quality metrics ROC AUC $0.975$ and PR AUC $0.0890$, making $1.18 \cdot 10^3$ correct predictions, while missing $2.09 \cdot 10^3$ earthquakes and making $192 \cdot 10^3$ false alarms. The baseline approach has similar ROC AUC $0.992$, number of correct predictions $1.19 \cdot 10^3$, and missing $2.07 \cdot 10^3$ earthquakes, but significantly worse PR AUC $0.00911$, and number of false alarms $1004 \cdot 10^3$.

According to an unofficial consensus, the birth of artificial intelligence as an independent research project can be dated to the summer of 1956, when John McCarthy at Dartmouth College, where he belonged to the Mathematical Department, was able to persuade the Rockefeller Foundation to finance an investigation " The study is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it". In addition to McCarthy (who was a professor at Stanford University until 2000 and is responsible for the coining of the term "artificial intelligence"), several other participants took part in the historical workshop at Dartmouth: Marvin Minsky (former professor at Stanford University), Claude Shannon (inventor of information theory); Herbert Simon (Nobel Prize winner in economics); Arthur Samuel (developer of the first chess computer program at world champion level); furthermore half a dozen experts from science and industry, who dreamed that it might be possible to produce a machine for coping with human tasks, which, according to the previous opinion, require intelligence. The Manifesto of Dartmouth (written at the dawn of the AI age) is both irritating and blurred. It is not clear whether the conference participants believed that one-day, machines would think or behave as if they could imagine. Both possible interpretations allow the word "simulate."

Back in August 2019, the BBC made some waves with the news that it was developing a voice assistant called Beeb, an English language "Alexa" of its own that could interact with and control its array of radio and TV services, and its on-demand catalogue, and able to understand the array of accents you find in across the BBC's national footprint to boot. Ten months on, it's releasing its first live version of the service in the form of a beta to a select group of early adopters: UK-based members of the Windows Insider Program, a beta-testing, bug-seeking, early-adopter group popular in the Windows community, with over 10 million users globally. The idea with the limited release beta -- according to Grace Boswood, COO of BBC Design and Engineering -- will be to get Insiders to try out various features and stress test Beeb in the early beta, while at the same time giving the BBC a trove of usage data that can help it continue to train Beeb further, ahead of a wider release. The BBC is not naming a date yet for the general release. When you are a member in the UK, you have to be using the latest release of Windows 10, and then you download Beeb BETA form the Windows App Store.)

Is it possible some instances of artificial intelligence are not as intelligent as we thought? Call it artificial artificial intelligence. A team of computer graduate students reports that a closer examination of several dozen information retrieval algorithms hailed as milestones in artificial research were in fact nowhere near as revolutionary as claimed. In fact, AI used in those algorithms were often merely minor tweaks of previously established routines. According to graduate student researcher Davis Blalock at the Massachusetts Institute of Technology, after his team examined 81 approaches to developing neural networks commonly believed to be superior to earlier efforts, the team could not confirm that any improvement, in fact, was ever achieved.

The Defense Department is seeking to adapt artificial intelligence technology it uses to track down terrorists with drones or predict when aircraft need maintenance for a new purpose: screening and testing novel coronavirus treatments and vaccines. The Pentagon plans to boost existing programs with money Congress provided under the virus-relief CARES Act for the "development of artificial intelligence-based models to rapidly screen, prioritize, and test Food and Drug Administration approved therapeutics for new COVID-19 drug candidates." The AI funds would also be tapped for human test trials for vaccines and antibody based treatments, according to the spending plan the department submitted to congressional panels. Dick Durbin (Ill.), the Senate's No. 2 Democrat and ranking member on the Appropriations Defense Subcommittee, pressed for the plan's release. While the amount of money the Pentagon wants to use on these programs is small---close to $1 million--it shows some of the department's urgency to apply new technology to choke off the pandemic.

Even the best inventions and intentions can result in unintended consequences. Email vastly improved many forms of communication and information sharing -- but it also begat spam, phishing, and an entire industry in cybersecurity. Social media connected billions of people and spread democratic ideals -- but it also wrought hacked accounts, stolen data, "fake news," and election meddling. The same is true with today's "advanced technologies" that promise to revolutionize information gathering, data analytics, workplace mobility, and much more during the coming decade. The ethical considerations and possible regulation of artificial intelligence (AI), machine learning, robotics, and other advanced technologies are playing catch-up once again.

Identifying different foods on a tray can be a non-trivial task (Image source: Chris A. Tweten on Unsplash) Unfortunately, we don't always have sufficient time to enjoy a leisurely lunchtime meal. Suppose some of your workmates invite you to join them for lunch at a local fast-food restaurant. You all typically have a limited amount of time for your lunchtime break and you need to use this time wisely and efficiently. First, you have to get to the restaurant, either by walking or perhaps by taking a short drive. When you reach the restaurant, you have to select your food, pay for it, and eat it.

One of the hottest topics in robotics is the field of soft robots, which utilizes squishy and flexible materials rather than traditional rigid materials. But soft robots have been limited due to their lack of good sensing. A good robotic gripper needs to feel what it is touching (tactile sensing), and it needs to sense the positions of its fingers (proprioception). Such sensing has been missing from most soft robots. In a new pair of papers, researchers from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) came up with new tools to let robots better perceive what they're interacting with: the ability to see and classify items, and a softer, delicate touch.