We've come a long way since the 70s, and computers today don't rape or mutilate people wearing black face masks. Instead, they enable the transition to a cashless economy, from which the poorest in society will be excluded (one cannot drop a credit card into a beggar's hat, and neither would anyone like to have their card swiped by a card reader on the street). AI systems assist judges in estimating the risk posed to society by particular offenders and decide on sentences or bail based on algorithms that are kept hidden from view. AI is at the forefront of environmental destruction, not only through the use of vast amounts of energy and carbon emissions for training computationally intensive models, but also by supporting research that will open up the Arctic to commercial exploitation, by enabling fossil-fuel companies to find new sources of oil, and by being one of the industries with the shortest time-to-obsolescence of its products. AI algorithms allow us to create new, terrifying weapons, come up with new types of terrorism, manipulate democratic processes, and endanger jobs on a global scale.

Existing exploration strategies in reinforcement learning (RL) often either ignore the history or feedback of search, or are complicated to implement. There is also a very limited literature showing their effectiveness over diverse domains. We propose an algorithm based on the idea of reannealing, that aims at encouraging exploration only when it is needed, for example, when the algorithm detects that the agent is stuck in a local optimum. The approach is simple to implement. We perform an illustrative case study showing that it has potential to both accelerate training and obtain a better policy.

This paper analyzes and compares different deep learning loss functions in the framework of multi-label remote sensing (RS) image scene classification problems. We consider seven loss functions: 1) cross-entropy loss; 2) focal loss; 3) weighted cross-entropy loss; 4) Hamming loss; 5) Huber loss; 6) ranking loss; and 7) sparseMax loss. All the considered loss functions are analyzed for the first time in RS. After a theoretical analysis, an experimental analysis is carried out to compare the considered loss functions in terms of their: 1) overall accuracy; 2) class imbalance awareness (for which the number of samples associated to each class significantly varies); 3) convexibility and differentiability; and 4) learning efficiency (i.e., convergence speed). On the basis of our analysis, some guidelines are derived for a proper selection of a loss function in multi-label RS scene classification problems.

Time-series forecasting is important for many applications. Forecasting models are usually trained using time-series data in a specific target task. However, sufficient data in the target task might be unavailable, which leads to performance degradation. In this paper, we propose a few-shot learning method that forecasts a future value of a time-series in a target task given a few time-series in the target task. Our model is trained using time-series data in multiple training tasks that are different from target tasks. Our model uses a few time-series to build a forecasting function based on a recurrent neural network with an attention mechanism. With the attention mechanism, we can retrieve useful patterns in a small number of time-series for the current situation. Our model is trained by minimizing an expected test error of forecasting next timestep values. We demonstrate the effectiveness of the proposed method using 90 time-series datasets.

In this article, we present a novel approach to multivariate probabilistic forecasting. Our approach is based on an extension of single-output quantile regression (QR) to multivariate-targets, called quantile surfaces (QS). QS uses a simple yet compelling idea of indexing observations of a probabilistic forecast through direction and vector length to estimate a central tendency. We extend the single-output QR technique to multivariate probabilistic targets. QS efficiently models dependencies in multivariate target variables and represents probability distributions through discrete quantile levels. Therefore, we present a novel two-stage process. In the first stage, we perform a deterministic point forecast (i.e., central tendency estimation). Subsequently, we model the prediction uncertainty using QS involving neural networks called quantile surface regression neural networks (QSNN). Additionally, we introduce new methods for efficient and straightforward evaluation of the reliability and sharpness of the issued probabilistic QS predictions. We complement this by the directional extension of the Continuous Ranked Probability Score (CRPS) score. Finally, we evaluate our novel approach on synthetic data and two currently researched real-world challenges in two different domains: First, probabilistic forecasting for renewable energy power generation, second, short-term cyclists trajectory forecasting for autonomously driving vehicles. Especially for the latter, our empirical results show that even a simple one-layer QSNN outperforms traditional parametric multivariate forecasting techniques, thus improving the state-of-the-art performance.

We propose a vision for directing research and education in the ICT field. Our Smart and Sustainable World vision targets at prosperity for the people and the planet through better awareness and control of both human-made and natural environment. The needs of the society, individuals, and industries are fulfilled with intelligent systems that sense their environment, make proactive decisions on actions advancing their goals, and perform the actions on the environment. We emphasize artificial intelligence, feedback loops, human acceptance and control, intelligent use of basic resources, performance parameters, mission-oriented interdisciplinary research, and a holistic systems view complementing the conventional analytical reductive view as a research paradigm especially for complex problems. To serve a broad audience, we explain these concepts and list the essential literature. We suggest planning research and education by specifying, in a step-wise manner, scenarios, performance criteria, system models, research problems and education content, resulting in common goals and a coherent project portfolio as well as education curricula. Research and education produce feedback to support evolutionary development and encourage creativity in research. Finally, we propose concrete actions for realizing this approach.

Let's go back to a simpler time. It is the early or late 90s. You are eight years old, waking up early to catch the latest action-filled episodes of your Saturday morning cartoons; TV shows that portray what technology may look like in the future. In Japan, popular anime shows like Outlaw Star, Mobile Suit Gundam, and Cowboy Bebop. These shows would pull viewers in, giving us a taste of the future for breakfast. They would show us worlds where humans and cyborgs were almost unidentifiable from each other, where trips to space were as simple as catching a bus, or where artificial intelligence and robotics were used to better humanity (and used for epic battles in space).

A way of monitoring household appliances by using machine learning to analyse vibrations on a wall or ceiling has been developed by researchers in the US. Their system could be used to create centralized smart home systems without the need for individual sensors in each object. What is more, the technology could help track energy use, identify electrical faults and even remind people to empty the dishwasher. "Recognizing home activities can help computers better understand human behaviours and needs, with the hope of developing a better human-machine interface," says team member and information scientist Cheng Zhang of Cornell University. The system, dubbed VibroSense, comprises two core parts: a laser Doppler vibrometer and a deep learning model, which is a type of machine learning system.

Robotics is a great way to introduce your child to a wide range of related brain-stimulating areas--coding, electronics, 3D printing, mechanical engineering--you name it. We live in an increasingly digitized, mechanized world, why not teach your child early on how to navigate our current world's STEM-heavy waters? Kids robotics kits teach children the value of hands-on learning and immerse them in the wide world of coding. Whether you're in the market for a battery operated, solar-powered, or hydraulic-run robot, we've got the perfect product for you. This kit includes the materials to build a robot on wheels that can turn around, follow a delineated path, and avoid obstacles.

Tijana Radivojevic (left) and Hector Garcia Martin working on mechanical and statistical modeling, data visualizations, and metabolic maps at the Agile BioFoundry last year. If you've eaten vegan burgers that taste like meat or used synthetic collagen in your beauty routine – both products that are "grown" in the lab – then you've benefited from synthetic biology. It's a field rife with potential, as it allows scientists to design biological systems to specification, such as engineering a microbe to produce a cancer-fighting agent. Yet conventional methods of bioengineering are slow and laborious, with trial and error being the main approach. Now scientists at the Department of Energy's Lawrence Berkeley National Laboratory (Berkeley Lab) have developed a new tool that adapts machine learning algorithms to the needs of synthetic biology to guide development systematically.