Sub-seasonal climate forecasting (SSF) focuses on predicting key climate variables such as temperature and precipitation in the 2-week to 2-month time scales. Skillful SSF would have immense societal value, in areas such as agricultural productivity, water resource management, transportation and aviation systems, and emergency planning for extreme weather events. However, SSF is considered more challenging than either weather prediction or even seasonal prediction. In this paper, we carefully study a variety of machine learning (ML) approaches for SSF over the US mainland. While atmosphere-land-ocean couplings and the limited amount of good quality data makes it hard to apply black-box ML naively, we show that with carefully constructed feature representations, even linear regression models, e.g., Lasso, can be made to perform well. Among a broad suite of 10 ML approaches considered, gradient boosting performs the best, and deep learning (DL) methods show some promise with careful architecture choices. Overall, suitable ML methods are able to outperform the climatological baseline, i.e., predictions based on the 30-year average at a given location and time. Further, based on studying feature importance, ocean (especially indices based on climatic oscillations such as El Nino) and land (soil moisture) covariates are found to be predictive, whereas atmospheric covariates are not considered helpful.

This paper studies how encouraging semantically-aligned features during deep neural network training can increase network robustness. Recent works observed that Adversarial Training leads to robust models, whose learnt features appear to correlate with human perception. Inspired by this connection from robustness to semantics, we study the complementary connection: from semantics to robustness. To do so, we provide a tight robustness certificate for distance-based classification models (clustering-based classifiers), which we leverage to propose ClusTR (Clustering Training for Robustness), a clustering-based and adversary-free training framework to learn robust models. Interestingly, ClusTR outperforms adversarially-trained networks by up to 4\% under strong PGD attacks. Moreover, it can be equipped with simple and fast adversarial training to improve the current state-of-the-art in robustness by 16\%-29\% on CIFAR10, SVHN, and CIFAR100.

The actor-critic (AC) algorithm is a popular method to find an optimal policy in reinforcement learning. In the infinite horizon scenario, the finite-sample convergence rate for the AC and natural actor-critic (NAC) algorithms has been established recently, but under independent and identically distributed (i.i.d.) sampling and single-sample update at each iteration. In contrast, this paper characterizes the convergence rate and sample complexity of AC and NAC under Markovian sampling, with mini-batch data for each iteration, and with actor having general policy class approximation. We show that the overall sample complexity for a mini-batch AC to attain an $\epsilon$-accurate stationary point improves the best known sample complexity of AC by an order of $\mathcal{O}(\epsilon^{-1}\log(1/\epsilon))$, and the overall sample complexity for a mini-batch NAC to attain an $\epsilon$-accurate globally optimal point improves the existing sample complexity of NAC by an order of $\mathcal{O}(\epsilon^{-2}/\log(1/\epsilon))$. Moreover, the sample complexity of AC and NAC characterized in this work outperforms that of policy gradient (PG) and natural policy gradient (NPG) by a factor of $\mathcal{O}((1-\gamma)^{-3})$ and $\mathcal{O}((1-\gamma)^{-4}\epsilon^{-2}/\log(1/\epsilon))$, respectively. This is the first theoretical study establishing that AC and NAC attain orderwise performance improvement over PG and NPG under infinite horizon due to the incorporation of critic.

We are often interested in identifying the feasible subset of a decision space under multiple constraints to permit effective design exploration. If determining feasibility required computationally expensive simulations, the cost of exploration would be prohibitive. Bayesian search is data-efficient for such problems: starting from a small dataset, the central concept is to use Bayesian models of constraints with an acquisition function to locate promising solutions that may improve predictions of feasibility when the dataset is augmented. At the end of this sequential active learning approach with a limited number of expensive evaluations, the models can accurately predict the feasibility of any solution obviating the need for full simulations. In this paper, we propose a novel acquisition function that combines the probability that a solution lies at the boundary between feasible and infeasible spaces (representing exploitation) and the entropy in predictions (representing exploration). Experiments confirmed the efficacy of the proposed function.

We define an evolving in time Bayesian neural network called a Hidden Markov neural network. The weights of a feed-forward neural network are modelled with the hidden states of a Hidden Markov model, whose observed process is given by the available data. A filtering algorithm is used to learn a variational approximation to the evolving in time posterior over the weights. Training is pursued through a sequential version of Bayes by Backprop Blundell et al. 2015, which is enriched with a stronger regularization technique called variational DropConnect. The experiments test variational DropConnect on MNIST and display the performance of Hidden Markov neural networks on time series.

A simulation created stunning sunsets from alien worlds across the solar system. The animation transports viewers to the surface of Venus, Mars, Uranus and Saturn's largest moon Titan, allowing them to witness the sun dip into the horizon. As a planet rotates away from the sun's light, photons scattered in different directions that produce an array of colors. The sunset on Uranus is a light shade of blue that fades into a royal blue with hints of turquoise, while Titan's starts as a vibrant yellow then shifts into a fiery red. The animation transports viewers to the surface of Venus, Mars, Uranus and Saturn's largest moon Titan, allowing them to witness the sun dip into the horizon.

A special effects and technology company in San Francisco is pitching a surprisingly lifelike animatronic dolphin to aquariums and marine parks in China to help them deal with the country's recent ban on wildlife trade. The dolphin was developed by Edge Innovations, a company founded by special effects veteran Walt Conti, who previously worked on The Abyss, Anaconda, Deep Blue Sea, The Perfect Storm, among many others. The current prototype was modeled after an adolescent bottlenose dolphin, weighs around 595 pounds, and can swim for 10 continuous hours on a single battery charge. Edge Innovations has developed a shockingly lifelike animatronic robot that it's pitching to marine parks and aquariums in China as a cheaper alternative to real dolphins The dolphin was designed to mimic the skeletal structure of a real dolphin, and uses internal bladders and weight deposits to further match a real dolphin's swimming movements - and its teeth have been given a light yellow staining for extra realism. The dolphin requires a human operator to swim and can't operate autonomously, according to a report in Gizmodo..

Everybody dances, every culture throughout history has danced. But our days of monopolizing the move busting market could soon be coming to an end, as Facebook AI has become the latest team to teach an AI to bop along to the beat. "In this work, we focus on designing interesting choreographies by combining the best of what humans are naturally good at – heuristics of'good' dance that an audience might find appealing – and what machines are good at – optimizing well-defined objective functions," the team wrote in a study published Tuesday. This isn't the first time we've tried to teach AIs to dance. In 2016, Swedish Choreographer Louise Crnkovic-Friis and her husband, Peltarion CEO Luka Crnkovic-Friis, trained a recurrent neural network, dubbed Chor-rnn, on 48 hours of Louise's movements.

Last week, the Supreme Court issued a surprising 6–3 decision barring hiring discrimination against LGBTQ people under Title VII of the Civil Rights Act, with conservative Justice Neil Gorsuch making the textualist case for this landmark protection. The unexpected outcome in Bostock v. Clayton County should provoke introspection among progressives in the legal community who have long been skeptical of textualism, offering a chance for them to fix chronic blind spots and strategic gaffes that have damaged the progressive judicial project. While it's clear that this ruling was a major victory for progressives, less apparent is how, going forward, progressive advocates, judges, and politicians should think and talk about statutory interpretation. Although brow-furrowing, that question is hugely important. As the late high priest of conservative textualism, Justice Antonin Scalia, pointed out: "By far the greatest part of what I and all federal judges do is interpret the meaning of federal statutes."

In a new letter, more than 1,600 Google workers are demanding that the company end its work with police departments across the country. "The past weeks have shown us that addressing racism is not merely an issue of words, but of actions taken to dismantle the actual structures that perpetuate it," the workers wrote in a draft of the to-be-released letter addressed to Google CEO Sundar Pichai and obtained by Mother Jones. The letter was signed by 1,670 employees, according to a screenshot that was shared with Mother Jones by someone with access to the signature list, and was organized by Googlers Against Racism, an advocacy group within the company. "While we as individuals hold difficult but necessary conversations with our family, friends and peers, we are also incredibly disappointed by our company's response," the letter continued, referencing Google's lip service to the Black Lives Matter movement. The letter also demands that the company "stop making our technology available to police forces."