The HapBlock program and its supplemental documents can be downloaded from the website http://www.cmb.usc.edu/~msms/HapBlock.

No abstract

Loyalty is a permanent concern to dictators. It is unclear, however, whether loyalty to a dictator assures the ruling group's cohesiveness. This study shows that authoritarian political elites under promotion pressure, while remaining loyal to their superior, also compete within factions to outrival their peers. Exploiting data on Chinese provincial leaders and local media reports on corruption investigations and industrial accidents (2000–2014), we find that Chinese elites promote negative news related to their co-faction peers as frequently as - or, depending on the measure, even more often than - they promote similar news regarding members of other factions. We also find that negative reports indeed reduce the promotion probability of reported cadres, while increasing that of reporting ones.

Deep reinforcement learning (DRL) has achieved super-human performance on complex video games (e.g., StarCraft II and Dota II). However, current DRL systems still suffer from challenges of multi-agent coordination, sparse rewards, stochastic environments, etc. In seeking to address these challenges, we employ a football video game, e.g., Google Research Football (GRF), as our testbed and develop an end-to-end learning-based AI system (denoted as TiKick 23 ) to complete this challenging task. In this work, we first generated a large replay dataset from the self-playing of single-agent experts, which are obtained from league training. We then developed a distributed learning system and new offline algorithms to learn a powerful multi-agent AI from the fixed single-agent dataset. To the best of our knowledge, Tikick is the first learning-based AI system that can take over the multi-agent Google Research Football full game, while previous work could either control a single agent or experiment on toy academic scenarios. Extensive experiments further show that our pre-trained model can accelerate the training process of the modern multi-agent algorithm and our method achieves state-of-theart performances on various academic scenarios. * Equal contribution 2 Codes can be found at https://github.com/TARTRL/TiKick.3 Videos available at https://sites.google.com/view/tikick.Preprint. Under review.

No abstract

No abstract

Quadcopter is an important way for the human to explore the physical world. The brain-computer interface (BCI) technology is used to control the quadcopter flight in order to help disabled persons communicate with the external world freely. In this study, a quadcopter control system using a hybrid BCI based on off-line optimization and enhanced human-machine interaction was designed to control the quadcopter flight in 3D physical space. The proposed system implemented the control of quadcopter moving up/down, forward/backward, left/right by six different SSVEP, and turning left/right by left-hand and right-hand motor imagery. Meanwhile, the optimization of the control system and the human-machine interaction enhancement improved practicability in real-time use. Five subjects participated in an on-line experiment to control the quadcopter flight in real-time. The average classification accuracy of EEG-based commands in the on-line experiment was 87.09±2.82% and information transfer rate (ITR) was 0.857±0.085 bits/min. The results demonstrated the feasibility of multidirectional control of quadcopter flight in 3D space by using hybrid BCI technology and revealed the practicality and operability of the hybrid BCI control system based on off-line optimization and human-machine interaction enhancement. INDEX TERMS Quadcopter control system, motor imagery, steady-state visual evoked potential, off-line optimization, human-machine interaction.

No abstract