This study is one of the very few investigating the dioxin body burden of a group of child-bearing-aged women at an electronic waste (e-waste) recycling site (Taizhou, Zhejiang Province) (24 ( 2.83 years of age, 40% were primiparae) and a reference site (Lin'an city, Zhejiang Province, about 245 km away from Taizhou) (24 ( 2.35 years of age, 100% were primiparae) in China. Five sets of samples (each set consisted of human milk, placenta, and hair) were collected from each site. Body burdens of people from the e-waste processing site (human milk, 21.02 ( 13.81 pg WHO-TEQ 1998 /g fat (World Health Organization toxic equivalency 1998); placenta, 31.15 ( 15.67 pg WHO-TEQ 1998 /g fat; hair, 33.82 ( 17.74 pg WHO-TEQ 1998 /g dry wt) showed significantly higher levels of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/ Fs) than those from the reference site (human milk, 9.35 ( 7.39 pg WHO-TEQ 1998 /g fat; placenta, 11.91 ( 7.05 pg WHO-TEQ 1998 /g fat; hair, 5.59 ( 4.36 pg WHO-TEQ 1998 /g dry wt) and were comparatively higher than other studies. The difference between the two sites was due to e-waste recycling operations, for example, open burning, which led to high background levels. Moreover, mothers from the e-waste recycling site consumed more foods of animal origin. The estimated daily intake of PCDD/Fs within 6 months by breastfed infants from the e-waste processing site was 2 times higher than that from the reference site. Both values exceeded the WHO tolerable daily intake for adults by at least 25 and 11 times, respectively. Our results implicated that e-waste recycling operations cause prominent PCDD/F levels in the environment and in humans. The elevated body burden may have health implications for the next generation.

No abstract

Acquiring mathematical skills is considered a key challenge for modern Artificial Intelligence systems. Inspired by the way humans discover numerical knowledge, here we introduce a deep reinforcement learning framework that allows to simulate how cognitive agents could gradually learn to solve arithmetic problems by interacting with a virtual abacus. The proposed model successfully learn to perform multi-digit additions and subtractions, achieving an error rate below 1% even when operands are much longer than those observed during training. We also compare the performance of learning agents receiving a different amount of explicit supervision, and we analyze the most common error patterns to better understand the limitations and biases resulting from our design choices.

With the development of science and technology, the intelligence of society is getting higher and higher. The development of safe and reliable intelligent driving has become one of the important topics in today’s era. As a high-tech complex integrating environmental awareness, planning decision-making, control execution and information interaction, driverless cars are increasingly becoming the needs of the times.[1] How to accurately extract effective information to provide safe and reliable control instructions for the next decision-making of vehicles in complex multi-source and heterogeneous environment is the technical basis of intelligent driving. In this paper, a certain amount of research has been done on lane change trajectory planning in different environments. In the aspect of static driving environment, a lane-changing trajectory planning method based on the fifth-order Bessel curve is planned. The physical problems are transformed into optimization problems of coordinate points of Bessel curve expression by planning the left and right boundaries of feasible region through lane-changing trajectory. [2]In the aspect of dynamic driving, based on the model predictive control theory, a lane-changing trajectory optimization method for multi-lane and multi-surrounding vehicles is proposed.