Despite the recent advances of deep reinforcement learning (DRL), agents trained by DRL tend to be brittle and sensitive to the training environment, especially in the multi-agent scenarios. In the multi-agent setting, a DRL agent’s policy can easily get stuck in a poor local optima w.r.t. its training partners – the learned policy may be only locally optimal to other agents’ current policies. In this paper, we focus on the problem of training robust DRL agents with continuous actions in the multi-agent learning setting so that the trained agents can still generalize when its opponents’ policies alter. To tackle this problem, we proposed a new algorithm, MiniMax Multi-agent Deep Deterministic Policy Gradient (M3DDPG) with the following contributions: (1) we introduce a minimax extension of the popular multi-agent deep deterministic policy gradient algorithm (MADDPG), for robust policy learning; (2) since the continuous action space leads to computational intractability in our minimax learning objective, we propose Multi-Agent Adversarial Learning (MAAL) to efficiently solve our proposed formulation. We empirically evaluate our M3DDPG algorithm in four mixed cooperative and competitive multi-agent environments and the agents trained by our method significantly outperforms existing baselines.

Three noncovalently fused-ring electron acceptors with good solubility, near-infrared absorption, and high electron mobility are precisely designed for high-efficiency OSCs.

Epidemiological studies have suggested inconclusive associations between 25-hydroxyvitamin D and total cancer incidence and mortality. The aim of this study was to quantitatively assess these associations by combining results from prospective cohort studies. A systematic literature search was implemented in PubMed and Scopus databases in April 2019. Comparing the highest with the lowest categories, the multivariate-adjusted relative risks (RRs) and the corresponding 95% confidence intervals (CIs) were pooled using a random-effects model. A trend estimation was performed using a two-stage, dose-response, meta-analysis method. Twenty-three independent prospective studies were included for data synthesis. Eight studies investigated the association between 25-hydroxyvitamin D and the risk of cancer incidence (7511 events and 70,018 participants), and the summary estimate showed that 25-hydroxyvitamin D is marginally associated with cancer risk (Summary RR = 0.86; 95% CI: 0.73, 1.02; I2 = 70.8%; p = 0.001). Sixteen studies investigated the association between 25-hydroxyvitamin D and the risk of cancer mortality (8729 events and 101,794 participants), and a higher 25-hydroxyvitamin D concentration was inversely associated with the risk of cancer mortality (Summary RR = 0.81; 95% CI: 0.71, 0.93; I2 = 48.8%, p = 0.012). Dose-response analysis indicated that the risk of cancer incidence was reduced by 7% (RRs = 0.93; 95% CI: 0.91, 0.96), and the risk of cancer mortality was reduced by 2% (RRs = 0.98; 95% CI: 0.97, 0.99), with each 20 nmol/L increment of 25-hydroxyvitamin D concentration. This meta-analysis provides evidence that a higher 25-hydroxyvitamin D concentration is associated with a lower cancer incidence and cancer mortality.

No abstract

Due to the aggressive biological behavior, lacking of specific targets and unconquered therapeutic resistance of Triple Negative Breast Cancer (TNBC), current therapeutic strategies were still limited. Combining multiple treatments have...