Search for: All records

In situ optical measurements during chemical vapor deposition processes can offer insight into the chemical reactions and electronic phenomena that are occurring during a process. However, the tooling to make these measurements can be complex, difficult to align, and may even require a redesign of the entire vacuum deposition chamber. Herein, we present a setup that allows for in situ optical measurements using vacuum-compatible fiber optics that only requires a singular conflat feedthrough, eliminating the need for optical viewports and alignment of external light sources/detectors. Proof of performance is shown with a neutral density filter and an exemplary application of vapor doping a conjugated polymer (poly-3-hexylthiophene) using vapor-phase TiCl4. 

Semantic segmentation of medical images is pivotal in applications like disease diagnosis and treatment planning. While deep learning automates this task effectively, it struggles in ultra low-data regimes for the scarcity of annotated segmentation masks. To address this, we propose a generative deep learning framework that produces high-quality image-mask pairs as auxiliary training data. Unlike traditional generative models that separate data generation from model training, ours uses multi-level optimization for end-to-end data generation. This allows segmentation performance to guide the generation process, producing data tailored to improve segmentation outcomes. Our method demonstrates strong generalization across 11 medical image segmentation tasks and 19 datasets, covering various diseases, organs, and modalities. It improves performance by 10–20% (absolute) in both same- and out-of-domain settings and requires 8–20 times less training data than existing approaches. This greatly enhances the feasibility and cost-effectiveness of deep learning in data-limited medical imaging scenarios. 

Bi-level optimization methods in machine learning are popularly effective in subdomains of neural architecture search, data reweighting, etc. However, most of these methods do not factor in variations in learning difficulty, which limits their performance in real-world applications. To address the above problems, we propose a framework that imitates the learning process of humans. In human learning, learners usually focus more on the topics where mistakes have been made in the past to deepen their understanding and master the knowledge. Inspired by this effective human learning technique, we propose a multilevel optimization framework, learning from mistakes (LFM), for machine learning. We formulate LFM as a three-stage optimization problem: 1) the learner learns, 2) the learner relearns based on the mistakes made before, and 3) the learner validates his learning. We develop an efficient algorithm to solve the optimization problem. We further apply our method to differentiable neural architecture search and data reweighting. Extensive experiments on CIFAR-10, CIFAR-100, ImageNet, and other related datasets powerfully demonstrate the effectiveness of our approach. The code of LFM is available at: https://github.com/importZL/LFM. 

BPM 2024 International Workshops, Krakow, Poland, September 1–6, 2024, Revised Selected Papers 

Abstract This study presents detailed time-integrated and time-resolved spectral analysis of the Fermi Gamma-ray Burst Monitor observations of the bright GRB 231129C. The results reveal its distinct spectral characteristics, featuring a hard low-energy spectral index (α) and soft high-energy spectral index (β), similar to GRB 090902B, suggesting a possible dominance of thermal emission. Further analysis indicates that 92% of the spectral indices exceed the synchrotron “line of death,” with the hardest index atα∼ +0.44. Simultaneously, 53% of the spectra can be well fitted by the nondissipative photosphere model, supporting a potential origin from a nondissipative photosphere. Additionally, we observe strong correlations between the spectral indexαand peak energyE_pwith flux. For theα−Frelationship, we employF=F₀e^{(3.00±0.10)α}to describe it, whereas theE_p−Frelationship requires a smoothly bending power-law function. Based on the framework proposed by Hascoët et al. and Gao & Zhang, the jet characteristics of this burst were studied, revealing that both methods support the suitability of a pure fireball model for this GRB at small initial jet radii.