Elon Musk is facing yet another lawsuit as shareholders of Tesla accuse the chief executive and his company of overstating the effectiveness and safety of their electric vehicles' autopilot and full self-driving technologies. Shareholders have alleged in the proposed class action lawsuit that Tesla defrauded them over four years with false and misleading statements that concealed how its technologies – suspected as a possible cause of multiple fatal crashes – "created a serious risk of accident and injury". The case was filed Monday in a San Francisco federal court. The case centers on the financial fallout of Tesla's failed autopilot features, citing when the company's share price fell after reports that the National Highway Traffic Safety Administration and the Securities and Exchange Commission had begun investigating the technologies. The share price also fell 5.7% on 16 February 2023 after NHTSA forced a recall of more than 362,000 Tesla vehicles equipped with full self-driving beta software because they could be unsafe around intersections. "As a result of defendants' wrongful acts and omissions, and the precipitous decline in the market value of the company's common stock, plaintiff and other class members have suffered significant losses and damages," the complaint said.

Over 5000 Dynamic World image are produced every day, whereas traditional approaches to building land cover data can take months or years to produce. As a result of leveraging a novel deep learning approach, based on Sentinel-2 Top of Atmosphere, Dynamic World offers global land cover updating every 2–5 days depending on location. A major benefit of an AI-powered approach is the model looks at an incoming Sentinel-2 satellite image and, for every pixel in the image, estimates the degree of tree cover, how built up a particular area is, or snow coverage if there's been a recent snowstorm, for example. As a result of the European Commission's Copernicus Program making European Space Agency Sentinel data freely and openly available, products like Dynamic World are able to offer 10m resolution land cover data. This is important because quantifying data in higher resolution produces more accurate results for what's really on the surface of the Earth.

Taipei, Taiwan – At opposite ends of Southeast Asia, researchers Pornampai Narenpitak and Heri Kuswanto are both working on the same problem: Is it possible to mimic the cooling effects of volcanic eruptions to halt global warming? Using computer modelling and analysis, Narenpitak and Kuswanto are separately studying whether shooting large quantities of sulphur dioxide into the earth's stratosphere could have a similar effect on global temperatures as the eruption of Indonesia's Mount Tambora in 1815. The eruption, the most powerful in recorded history, spewed an estimated 150 cubic kilometres (150,000 gigalitres) of exploded rock and ash into the air, causing global temperatures to fall as much as 3 degrees Celsius (5.4 degrees Fahrenheit) in what became known as the "year without a summer". Stratospheric aerosol injection is among a number of nascent – and controversial – technologies in the field of solar geoengineering (SRM) that have been touted as potential solutions to mitigating the effects of climate change. Other proposed strategies include brightening marine clouds to reflect the sun or breaking up cirrus clouds that capture heat.

China led the Artificial Intelligence (AI) race a few years ago with abundant data, innovative entrepreneurs, and supportive policies. However, today the country has fallen behind in tech innovation, lagging far behind the United States. The ChatGPT, an advanced experimental chatbot, created by the American startup OpenAI with the help of Microsoft, is leaving China's tech entrepreneurs shocked and demoralized. Many are asking fundamental questions about China's innovation environment, with some suggesting that censorship, geopolitical tensions, and government control of the private sector have made China less innovation-friendly. The Chinese government is notorious for censorship, and its obsession with controlling online content is perhaps its most significant obstacle to technological advancements.

Modeling complex physical dynamics is a fundamental task in science and engineering. Traditional physics-based models are sample efficient, and interpretable but often rely on rigid assumptions. Furthermore, direct numerical approximation is usually computationally intensive, requiring significant computational resources and expertise, and many real-world systems do not have fully-known governing laws. While deep learning (DL) provides novel alternatives for efficiently recognizing complex patterns and emulating nonlinear dynamics, its predictions do not necessarily obey the governing laws of physical systems, nor do they generalize well across different systems. Thus, the study of physics-guided DL emerged and has gained great progress. Physics-guided DL aims to take the best from both physics-based modeling and state-of-the-art DL models to better solve scientific problems. In this paper, we provide a structured overview of existing methodologies of integrating prior physical knowledge or physics-based modeling into DL, with a special emphasis on learning dynamical systems. We also discuss the fundamental challenges and emerging opportunities in the area.

Pre-training representations (a.k.a. foundation models) has recently become a prevalent learning paradigm, where one first pre-trains a representation using large-scale unlabeled data, and then learns simple predictors on top of the representation using small labeled data from the downstream tasks. There are two key desiderata for the representation: label efficiency (the ability to learn an accurate classifier on top of the representation with a small amount of labeled data) and universality (usefulness across a wide range of downstream tasks). In this paper, we focus on one of the most popular instantiations of this paradigm: contrastive learning with linear probing, i.e., learning a linear predictor on the representation pre-trained by contrastive learning. We show that there exists a trade-off between the two desiderata so that one may not be able to achieve both simultaneously. Specifically, we provide analysis using a theoretical data model and show that, while more diverse pre-training data result in more diverse features for different tasks (improving universality), it puts less emphasis on task-specific features, giving rise to larger sample complexity for down-stream supervised tasks, and thus worse prediction performance. Guided by this analysis, we propose a contrastive regularization method to improve the trade-off. We validate our analysis and method empirically with systematic experiments using real-world datasets and foundation models.

More than half of all fatalities on U.S. highways occur due to roadway departure (RwD) each year. Previous research has explored various risk factors that contribute to RwD crashes, however, a comprehensive investigation considering the effect of lighting conditions has been insufficiently addressed. Using the Louisiana Department of Transportation and Development crash database, fatal and injury RwD crashes occurring on rural two-lane (R2L) highways between 2008-2017 were analyzed based on daylight and dark (with/without streetlight). This research employed a safe system approach to explore meaningful complex interactions among multidimensional crash risk factors. To accomplish this, an unsupervised data mining algorithm association rules mining (ARM) was utilized. Based on the generated rules, the findings reveal several interesting crash patterns in the daylight, dark-with-streetlight, and dark-no-streetlight, emphasizing the importance of investigating RwD crash patterns depending on the lighting conditions. In daylight, fatal RwD crashes are associated with cloudy weather conditions, distracted drivers, standing water on the roadway, no seat belt use, and construction zones. In dark lighting conditions (with/without streetlight), the majority of the RwD crashes are associated with alcohol/drug involvement, young drivers (15-24 years), driver condition (e.g., inattentive, distracted, illness/fatigued/asleep) and colliding with animal (s). The findings reveal how certain driver behavior patterns are connected to RwD crashes, such as a strong association between alcohol/drug intoxication and no seat belt usage in the dark-no-streetlight condition. Based on the identified crash patterns and behavioral characteristics under different lighting conditions, the findings could aid researchers and safety specialists in developing the most effective RwD crash mitigation strategies.

Because of the limitations of autonomous driving technologies, teleoperation is widely used in dangerous environments such as military operations. However, the teleoperated driving performance depends considerably on the driver's skill level. Moreover, unskilled drivers need extensive training time for teleoperations in unusual and harsh environments. To address this problem, we propose a novel denoising-based driver assistance method, namely GoonDAE, for real-time teleoperated off-road driving. The unskilled driver control input is assumed to be the same as the skilled driver control input but with noise. We designed a skip-connected long short-term memory (LSTM)-based denoising autoencoder (DAE) model to assist the unskilled driver control input by denoising. The proposed GoonDAE was trained with skilled driver control input and sensor data collected from our simulated off-road driving environment. To evaluate GoonDAE, we conducted an experiment with unskilled drivers in the simulated environment. The results revealed that the proposed system considerably enhanced driving performance in terms of driving stability.

The optimization of electrical circuits is a difficult and time-consuming process performed by experts, but also increasingly by sophisticated algorithms. In this paper, a reinforcement learning (RL) approach is adapted to optimize a LLC converter at multiple operation points corresponding to different output powers at high converter efficiency at different switching frequencies. During a training period, the RL agent learns a problem specific optimization policy enabling optimizations for any objective and boundary condition within a pre-defined range. The results show, that the trained RL agent is able to solve new optimization problems based on LLC converter simulations using Fundamental Harmonic Approximation (FHA) within 50 tuning steps for two operation points with power efficiencies greater than 90%. Therefore, this AI technique provides the potential to augment expert-driven design processes with data-driven strategy extraction in the field of power electronics and beyond.

Oversmoothing is a central challenge of building more powerful Graph Neural Networks (GNNs). While previous works have only demonstrated that oversmoothing is inevitable when the number of graph convolutions tends to infinity, in this paper, we precisely characterize the mechanism behind the phenomenon via a non-asymptotic analysis. Specifically, we distinguish between two different effects when applying graph convolutions -- an undesirable mixing effect that homogenizes node representations in different classes, and a desirable denoising effect that homogenizes node representations in the same class. By quantifying these two effects on random graphs sampled from the Contextual Stochastic Block Model (CSBM), we show that oversmoothing happens once the mixing effect starts to dominate the denoising effect, and the number of layers required for this transition is $O(\log N/\log (\log N))$ for sufficiently dense graphs with $N$ nodes. We also extend our analysis to study the effects of Personalized PageRank (PPR), or equivalently, the effects of initial residual connections on oversmoothing. Our results suggest that while PPR mitigates oversmoothing at deeper layers, PPR-based architectures still achieve their best performance at a shallow depth and are outperformed by the graph convolution approach on certain graphs. Finally, we support our theoretical results with numerical experiments, which further suggest that the oversmoothing phenomenon observed in practice can be magnified by the difficulty of optimizing deep GNN models.