State density distribution, in contrast to worst-case reachability, can be leveraged for safety-related problems to better quantify the likelihood of the risk for potentially hazardous situations. In this work, we propose a data-driven method to comp
ute the density distribution of reachable states for nonlinear and even black-box systems. Our semi-supervised approach learns system dynamics and the state density jointly from trajectory data, guided by the fact that the state density evolution follows the Liouville partial differential equation. With the help of neural network reachability tools, our approach can estimate the set of all possible future states as well as their density. Moreover, we could perform online safety verification with probability ranges for unsafe behaviors to occur. We use an extensive set of experiments to show that our learned solution can produce a much more accurate estimate on density distribution, and can quantify risks less conservatively and flexibly comparing with worst-case analysis.
Reactive and safe agent modelings are important for nowadays traffic simulator designs and safe planning applications. In this work, we proposed a reactive agent model which can ensure safety without comprising the original purposes, by learning only
high-level decisions from expert data and a low-level decentralized controller guided by the jointly learned decentralized barrier certificates. Empirical results show that our learned road user simulation models can achieve a significant improvement in safety comparing to state-of-the-art imitation learning and pure control-based methods, while being similar to human agents by having smaller errors to the expert data. Moreover, our learned reactive agents are shown to generalize better to unseen traffic conditions, and react better to other road users and therefore can help understand challenging planning problems pragmatically.
We report the first dark matter search results using the commissioning data from PandaX-4T. Using a time projection chamber with 3.7-tonne of liquid xenon target and an exposure of 0.63~tonne$cdot$year, 1058 candidate events are identified within an
approximate electron equivalent energy window between 1 and 30 keV. No significant excess over background is observed. Our data set a stringent limit to the dark matter-nucleon spin-independent interactions, with a lowest excluded cross section (90% C.L.) of $3.3times10^{-47} $cm$^2$ at a dark matter mass of 30 GeV/$c^2$.
This paper focuses on building semantic maps, containing object poses and shapes, using a monocular camera. This is an important problem because robots need rich understanding of geometry and context if they are to shape the future of transportation,
construction, and agriculture. Our contribution is an instance-specific mesh model of object shape that can be optimized online based on semantic information extracted from camera images. Multi-view constraints on the object shape are obtained by detecting objects and extracting category-specific keypoints and segmentation masks. We show that the errors between projections of the mesh model and the observed keypoints and masks can be differentiated in order to obtain accurate instance-specific object shapes. We evaluate the performance of the proposed approach in simulation and on the KITTI dataset by building maps of car poses and shapes.
Temporal modelling is the key for efficient video action recognition. While understanding temporal information can improve recognition accuracy for dynamic actions, removing temporal redundancy and reusing past features can significantly save computa
tion leading to efficient action recognition. In this paper, we introduce an adaptive temporal fusion network, called AdaFuse, that dynamically fuses channels from current and past feature maps for strong temporal modelling. Specifically, the necessary information from the historical convolution feature maps is fused with current pruned feature maps with the goal of improving both recognition accuracy and efficiency. In addition, we use a skipping operation to further reduce the computation cost of action recognition. Extensive experiments on Something V1 & V2, Jester and Mini-Kinetics show that our approach can achieve about 40% computation savings with comparable accuracy to state-of-the-art methods. The project page can be found at https://mengyuest.github.io/AdaFuse/
Action recognition is an open and challenging problem in computer vision. While current state-of-the-art models offer excellent recognition results, their computational expense limits their impact for many real-world applications. In this paper, we p
ropose a novel approach, called AR-Net (Adaptive Resolution Network), that selects on-the-fly the optimal resolution for each frame conditioned on the input for efficient action recognition in long untrimmed videos. Specifically, given a video frame, a policy network is used to decide what input resolution should be used for processing by the action recognition model, with the goal of improving both accuracy and efficiency. We efficiently train the policy network jointly with the recognition model using standard back-propagation. Extensive experiments on several challenging action recognition benchmark datasets well demonstrate the efficacy of our proposed approach over state-of-the-art methods. The project page can be found at https://mengyuest.github.io/AR-Net
The crystal structure of iron in the Earths inner core remains debated. Most recent experiments suggest a hexagonal-close-packed (hcp) phase. In simulations, it has been generally agreed that the hcp-Fe is stable at inner core pressures and relativel
y low temperatures. At high temperatures, however, several studies suggest a body-centered-cubic (bcc) phase at the inner core condition. We have examined the crystal structure of iron at high pressures over 2 million atmospheres (>200GPa) and at high temperatures over 5000 kelvin in a laser-heated diamond cell using microstructure analysis combined with $textit{in-situ}$ x-ray diffraction. Experimental evidence shows a bcc-Fe appearing at core pressures and high temperatures, with an hcp-bcc transition line in pressure-temperature space from about 95$pm$2GPa and 2986$pm$79K to at least 222$pm$6GPa and 4192$pm$104K. The trend of the stability field implies a stable bcc-Fe at the Earths inner core condition, with implications including a strong candidate for explaining the seismic anisotropy of the Earths inner core.
