Category Archives: Publication

NEW PAPER: “Revisiting the Edge of Chaos: Again?”

C. Teuscher. “Revisiting the Edge of Chaos: Again?” Biosystems 18:104693, 2022, https://doi.org/10.1016/j.biosystems.2022.104693

Abstract: Does biological computation happen at some sort of “edge of chaos”, a dynamical regime somewhere between order and chaos? And if so, is this a fundamental principle that underlies self-organization, evolution, and complex natural and artificial systems that are subjected to adaptation? In this article, we will review the literature on the fundamental principles of computation in natural and artificial systems at the “edge of chaos”. The term was coined by Norman Packard in the late 1980s. Since then, the concept of “adaptation to the edge of chaos” was demonstrated and investigated in many fields where both simple and complex systems receive some sort of feedback. Besides reviewing both historic and recent literature, we will also review critical voices of the concept.

Nithya, Nancy, and Ben present posters at Student Research Symposium

New Science article: Reconfigurable perovskite nickelate electronics for artificial intelligence

Our new Science article is out:

H.-T. Zhang and T. J. Park and A. N. M. N. Islam and D. S. J. Tran and S. Manna and Q. Wang and S. Mondal and H. Yu and S. Banik and S. Cheng and H. Zhou and S. Gamage and S. Mahapatra and Y. Zhu and Y. Abate and N. Jiang and S. K. R. S. Sankaranarayanan and A. Sengupta and C. Teuscher and S. Ramanathan. Reconfigurable perovskite nickelate electronics for artificial intelligence. Science, 375(6580): 533-539, 2022. https://doi.org/10.1126/science.abj7943

“Having all the core functionality required for neuromorphic computing in one type of a device could offer dramatic improvements to emerging computing architectures and brain-inspired hardware for artificial intelligence. Zhang et al. showed that proton-doped perovskite neodymium nickelate (NdNiO3) could be reconfigured at room temperature by simple electrical pulses to generate the different functions of neuron, synapse, resistor, and capacitor (see the Perspective by John). The authors designed a prototype experimental network that not only demonstrated electrical reconfiguration of the device, but also showed that such dynamic networks enabled a better approximation of the dataset for incremental learning scenarios compared with static networks.” —YS

Science commentary: https://doi.org/10.1126/science.abn6196

PSU press release: https://www.pdx.edu/news/new-ai-research-gives-existing-systems-versatility-growth-and-lifelong-learning

Other press coverage:

New Paper: Proximal Policy Optimization for Radiation Source Search

Proctor, P.; Teuscher, C.; Hecht, A.; Osiński, M. Proximal Policy Optimization for Radiation Source Search. Journal of Nuclear Engineering, 2:368-397, 2021. https://doi.org/10.3390/jne2040029

Rapid search and localization for nuclear sources can be an important aspect in preventing human harm from illicit material in dirty bombs or from contamination. In the case of a single mobile radiation detector, there are numerous challenges to overcome such as weak source intensity, multiple sources, background radiation, and the presence of obstructions, i.e., a non-convex environment. In this work, we investigate the sequential decision making capability of deep reinforcement learning in the nuclear source search context. A novel neural network architecture (RAD-A2C) based on the advantage actor critic (A2C) framework and a particle filter gated recurrent unit for localization is proposed. Performance is studied in a randomized 20×20 m convex and non-convex simulation environment across a range of signal-to-noise ratio (SNR)s for a single detector and single source. RAD-A2C performance is compared to both an information-driven controller that uses a bootstrap particle filter and to a gradient search (GS) algorithm. We find that the RAD-A2C has comparable performance to the information-driven controller across SNR in a convex environment. The RAD-A2C far outperforms the GS algorithm in the non-convex environment with greater than 95% median completion rate for up to seven obstructions.