DNA Family: Boosting Weight-Sharing NAS With Block-Wise Supervisions.

Neural Architecture Search (NAS), aiming at automatically designing neural architectures by machines, has been considered a key step toward automatic machine learning. One notable NAS branch is the weight-sharing NAS, which significantly improves search efficiency and allows NAS algorithms to run on ordinary computers. Despite receiving high expectations, this category of methods suffers from low search effectiveness. By employing a generalization boundedness tool, we demonstrate that the devil behind this drawback is the untrustworthy architecture rating with the oversized search space of the possible architectures. Addressing this problem, we modularize a large search space into blocks with small search spaces and develop a family of models with the distilling neural architecture (DNA) techniques. These proposed models, namely a DNA family, are capable of resolving multiple dilemmas of the weight-sharing NAS, such as scalability, efficiency, and multi-modal compatibility. Our proposed DNA models can rate all architecture candidates, as opposed to previous works that can only access a sub- search space using heuristic algorithms. Moreover, under a certain computational complexity constraint, our method can seek architectures with different depths and widths. Extensive experimental evaluations show that our models achieve state-of-the-art top-1 accuracy of 78.9% and 83.6% on ImageNet for a mobile convolutional network and a small vision transformer, respectively. Additionally, we provide in-depth empirical analysis and insights into neural architecture ratings.

Assessment of the Adverse Health Effects of Aflatoxin Exposure from Unpackaged Peanut Oil in Guangdong, China.

Aflatoxins are liver carcinogens and are common contaminants in unpackaged peanut (UPP) oil. However, the health risks associated with consuming aflatoxins in UPP oil remain unclear. In this study, aflatoxin contamination in 143 UPP oil samples from Guangdong Province were assessed via liquid chromatography-tandem mass spectrometry (LC-MS). We also recruited 168 human subjects, who consumed this oil, to measure their liver functions and lipid metabolism status. Aflatoxin B1 (AFB1) was detected in 79.72% of the UPP oil samples, with levels ranging from 0.02 to 174.13 µg/kg. The average daily human intake of AFB1 from UPP oil was 3.14 ng/kg·bw/day; therefore, the incidence of liver cancer, caused by intake of 1 ng/kg·bw/day AFB1, was estimated to be 5.32 cases out of every 100,000 persons per year. Meanwhile, Hepatitis B virus (HBV) infection and AFB1 exposure exerted a synergistic effect to cause liver dysfunction. In addition, the triglycerides (TG) abnormal rate was statistically significant when using AFB1 to estimate daily intake (EDI) quartile spacing grouping (p = 0.011). In conclusion, high aflatoxin exposure may exacerbate the harmful effects of HBV infection on liver function. Contamination of UPP oil with aflatoxins in Guangdong urgently requires more attention, and public health management of the consumer population is urgently required.