Adversarial attacks against conventional Deep Learning (DL) systems and algorithms have been widely studied, and various defenses were proposed. However, the possibility and feasibility of such attacks against Deep Reinforcement Learning (DRL) are less explored. As DRL has achieved great success in various complex tasks, designing effective adversarial attacks is an indispensable prerequisite towards building robust DRL algorithms. In this paper, we introduce two novel adversarial attack techniques to stealthily and efficiently attack the DRL agents. These two techniques enable an adversary to inject adversarial samples in a minimal set of critical moments while causing the most severe damage to the agent. The first technique is the critical point attack: the adversary builds a model to predict the future environmental states and agent's actions, assesses the damage of each possible attack strategy, and selects the optimal one. The second technique is the antagonist attack: the adversary automatically learns a domain-agnostic model to discover the critical moments of attacking the agent in an episode. Experimental results demonstrate the effectiveness of our techniques. Specifically, to successfully attack the DRL agent, our critical point technique only requires 1 (TORCS) or 2 (Atari Pong and Breakout) steps, and the antagonist technique needs fewer than 5 steps (4 Mujoco tasks), which are significant improvements over state-of-the-art methods.
A high responsivity and controllable recovery ultraviolet (UV) photodetector based on a tungsten oxide (WO3) gate AlGaN/GaN heterostructure with an integrated micro-heater is reported for the first time.
We developed an AlGaN/GaN high electron mobility transistor (HEMT) sensor with a tungsten trioxide (WO 3 ) nano-film modified gate for nitrogen dioxide (NO 2 ) detection. The device has a suspended circular membrane structure and an integrated micro-heater. The thermal characteristic of the Platinum (Pt) micro-heater and the HEMT self-heating are studied and modeled. A significant detection is observed for exposure to a low concentration of 100 ppb NO 2 /N 2 at ∼300 • C. For a 1 ppm NO 2 gas, a high sensitivity of 1.1% with a response (recovery) time of 88 second (132 second) is obtained. The effects of relative humidity and temperature on the gas sensor response properties in air are also studied. Based on the excellent sensing performance and inherent advantages of low power consumption, the investigated sensor provides a viable alternative high performance NO 2 sensing applications. It is suitable for continuous environmental monitoring system or high temperature applications.
Laboratory experiments have been carried out to model the magnetic reconnection process in a solar flare with powerful lasers. Relativistic electrons with energy up to MeV are detected along the magnetic separatrices bounding the reconnection outflow, which exhibit a kappa-like distribution with an effective temperature of ∼10 9 K. The acceleration of non-thermal electrons is found more efficient in the case with a guide magnetic field (a component of magnetic field along the reconnection-induced electric field) than that in the case without a guide field. Hardening of the spectrum at energies ≥ 500 keV is observed in both cases, which remarkably resembles the hardennings of hard X-ray and γ-ray spectra observed in many solar flares. This supports a recent proposal that the hardening in the hard X-ray and γ-ray emissions of solar flares is due to a hardening of the source-electron spectrum. We also performed numerical simulations that help examine behaviors of electrons in the reconnection process with the electromagnetic field configurations occurring in the experiments. Trajectories of non-thermal electrons observed in the experiments were well duplicated in the simulations. Our numerical simulations generally reproduce the electron energy spectrum as well, except the hardening of the electron spectrum. This suggests that other mechanisms such as shock and/or turbulence may play an important role in productions of the observed energetic electron.
A suspended AlGaN/GaN high electron mobility transistor (HEMT) sensor with a tungsten trioxide (WO 3 ) nanofilm modified gate was microfabricated and characterized for ppm-level acetone gas detection. The sensor featured a suspended circular membrane structure and an integrated microheater to select the optimum working temperature. High working temperature (300 • C) increased the sensitivity to up to 25.7% and drain current change I DS to 0.31 mA for 1000-ppm acetone in dry air. The transient characteristics of the sensor exhibited stable operation and good repeatability at different temperatures. For 1000-ppm acetone concentration, the measured response and recovery times reduced from 148 and 656 to 48 and 320 s as the temperature increased from 210 • C to 300 • C. The sensitivity to 1000-ppm acetone gas was significantly greater than the sensitivity to ethanol, ammonia, and CO gases, showing low cross-sensitivity. These results demonstrate a promising step toward the realization of an acetone sensor based on the suspended AlGaN/GaN HEMTs.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.