Cutting to the chase with warm-start contextual bandits

Knowledge and Information Systems - Multi-armed bandits achieve excellent long-term performance in practice and sublinear cumulative regret in theory. However, a real-world limitation of bandit...     

Tailoring 3D Printed Micro-Structured Carbons for Adsorption

The manufacture of tailored carbon-based adsorbent structures with exceptionally low-pressure drops and improved kinetics using stereolithographic 3D printing is presented. Adsorbent structures are printed from commercial resins with square, circular, and hexagonal cross-sectional microchannels. These structures can reduce energy use by 50–95% compared to conventional carbon-packed beds. The activated 3D printed carbon achieves Brunauer–Emmett–Teller surface areas over 1000 m² g⁻¹ and shows outstanding butane adsorption capacities, over twice the capacity of a commercial carbon and a comparable capacity to phenolic-based carbons. The structures also show excellent uptakes of cyclohexane, up to 0.62 g g⁻¹ in a saturated feed. The introduction of complex axial geometries including spirals and chevrons enable superior adsorption kinetics and premature breakthrough of contaminants at high gas flow rates. These results demonstrate the success of intelligent manufacturing of low-pressure drop, high-capacity micro-structured adsorbents, allowing for the development of gas separation technologies for applications such as greenhouse gas removal and respiratory protection.     

Neural network-based optical flow versus traditional optical flow techniques with thermal aerial imaging in real-world settings

The study explores the feasibility of optical flow-based neural network from real-world thermal aerial imagery. While traditional optical flow techniques have shown adequate performance, sparse techniques do not work well during cold-soaked low-contrast conditions, and dense algorithms are more accurate in low-contrast conditions but suffer from the aperture problem in some scenes. On the other hand, optical flow from convolutional neural networks has demonstrated good performance with strong generalization from several synthetic public data set benchmarks. Ground truth was generated from real-world thermal data estimated with traditional dense optical flow techniques. The state-of-the-art Recurrent All-Pairs Field Transform for the Optical Flow model was trained with both color synthetic data and the captured real-world thermal data across various thermal contrast conditions. The results showed strong performance of the deep-learning network against established sparse and dense optical flow techniques in various environments and weather conditions, at the cost of higher computational demand.