Zi He

Feng Xue, Zi He, Chuanlong Xie et al.

Out-of-Distribution (OOD) detection, i.e., identifying whether an input is sampled from a novel distribution other than the training distribution, is a critical task for safely deploying machine learning systems in the open world. Recently, post hoc detection utilizing pre-trained models has shown promising performance and can be scaled to large-scale problems. This advance raises a natural question: Can we leverage the diversity of multiple pre-trained models to improve the performance of post hoc detection methods? In this work, we propose a detection enhancement method by ensembling multiple detection decisions derived from a zoo of pre-trained models. Our approach uses the p-value instead of the commonly used hard threshold and leverages a fundamental framework of multiple hypothesis testing to control the true positive rate of In-Distribution (ID) data. We focus on the usage of model zoos and provide systematic empirical comparisons with current state-of-the-art methods on various OOD detection benchmarks. The proposed ensemble scheme shows consistent improvement compared to single-model detectors and significantly outperforms the current competitive methods. Our method substantially improves the relative performance by 65.40% and 26.96% on the CIFAR10 and ImageNet benchmarks.

Zi He, Ji Hong Gu, Wei E. I. Sha et al.

Metallic nanoparticles (NPs) support localized surface plasmon resonances (LSPRs), which enable to concentrate sunlight at the active layer of solar cells. However, full-wave modeling of the plasmonic solar cells faces great challenges in terms of huge computational workload and bad matrix condition. It is tremendously difficult to accurately and efficiently simulate near-field multiple scattering effects from plasmonic NPs embedded into solar cells. In this work, a preconditioned volume integral equation (VIE) is proposed to model plasmonic organic solar cells (OSCs). The diagonal block preconditioner is applied to different material domains of the device structure. As a result, better convergence and higher computing efficiency are achieved. Moreover, the calculation is further accelerated by two-dimensional periodic Green's functions. Using the proposed method, the dependences of optical absorption on the wavelengths and incident angles are investigated. Angular responses of the plasmonic OSCs show the super-Lambertian absorption on the plasmon resonance but near-Lambertian absorption off the plasmon resonance. The volumetric method of moments and explored physical understanding are of great help to investigate the optical responses of OSCs.