Search

According to optical diffraction limit, the photoresponsity of nanowire (NW)-based photodetector exponentially decreases when its NW diameter reduces to the range of deep subwavelength. In this paper, we demonstrate a photoresponse-enhanced method of the deep-subwavelength GaAs NW photodetector by using a plasmon-driven dipole antenna. Considering that the enhancement is extremely influenced by the shape and size of antenna, the structure of antenna is optimized by finite difference time domain (FDTD) solutions. The optimal structure of antenna optimizes the responsivity-enhanced factors to 1123.3 and 224.7 in NW photodetectors with NW diameters of 20 nm and 60 nm, respectively. This photoresponse-enhanced method is promising for easy-integration high-performance nanoscale photodetectors.

The formation of public opinions is a complex phenomenon that revolves around the aggregation of individuals’ beliefs. To accurately capture this phenomenon, one needs to build links from individualistic experiences to personal beliefs, which evolve in a social space through information exchange and belief revision. Despite many efforts to model opinion dynamics, the role of personal experiences and beliefs has often been overlooked. In this paper, we address this issue and propose an agent-based model in a social network. We explicitly model belief acquisition as a learning process from experiences that take the form of local data sets. Agents interact through a social network and update their beliefs based on how accurately the belief reflects experiences. Through iterations of interactions, the agents are able to arrive at a unified belief. We then focus on the accuracy of the personal beliefs during their evolution and the impact of the social network structure. On a micro-level, we investigate positional attributes such as the centrality of nodes that affect belief accuracy. On a macro-level, we investigate structural features that affect the overall performance. We then investigate a method to intervene in opinion formation through expert agents. Experiments are performed on real-world and synthetic data sets, which validate a number of important structural insights.

The transmission characters of a narrow slit were an interesting topic for a long time. In this paper, the transmission characters of an air slit carved form a metal film was discussed rigorously. The narrow slit was treated to be a three-layer-waveguide with two complex refractive index claddings. The transmission character eigen equations were derived by solving analytically from Maxwells equations and by matching the boundary conditions at the interfaces. The electromagnetic fields in both sides of the claddings and inside of the air slit were obtained. For a given example, numerical simulations were presented. The field distribution and the longitude phase constants at different wavelengths and with different width of the slits were calculated. Five modes were supported, and one surface Plasmon mode was attributed and four guided modes were attributed from their transmission characters. The p-polarized surface Plasmon mode might play an important role in the extraordinary optical transmission of light through arrayed subwavelength holes.

Ensemble learning aims to train and combine multiple base learners in the hope of improving the overall performance. Existing ensemble algorithms rely on a centralized framework where each base learner has access to the entire training dataset. We combine the technology of blockchains which is mainly used for data validation with ensemble learning and propose a decentralized framework where data are distributed among multiple base learners, who exchange their respective data to improve the collective predictive abilities. We develop two realizations of this framework, based on static and dynamic decision trees, respectively. We evaluate our methods over 20 real-world datasets and compare them against other centralized ensemble methods. Experimental results show that the proposed method obtains improved accuracy scores through sample exchange and achieves competitive performance with state-of-the-art ensemble methods whereas the base learners store only a small fraction of the samples.

We demonstrate a subwavelength GaAs nanowire photodetector with a half-wave Hertz dipole antenna. An enhancement by a factor of 100 in the responsivity is obtained due to the antenna resonance.

A general game playing (GGP) system aims to play previously unknown board games with changeable rules without human intervention. Taking changeable game rules into consideration, a game description language presents formal descriptions of a game. Based on this description, legal moves can be automatically generated so that each player in GGP system only needs to solve the problems of searching and learning for playing well. Traditional search methods demand the player to compute all legal moves, which can be very time consuming. In GGP, the coordinate of cells in the game board is very important in board game rules. Thus we address the relationship among cell coordinates. Borrowing an idea from rule learning to prune the board game search tree, we propose a new search optimization method to reduce running time when searching a large search space. We further prove that this method can effectively improve the searching efficiency through a comparative experiment with Gomoku game in GGP system.

The extent to which a node occupies a central position in a social network amounts to a crucial indicator of personal influence. Few works address the mechanisms that would allow an individual to integrate into a network, and even fewer examine the correlation between this process and various notions of centrality. In this paper, we tackle this problem by focusing on the process in which a newcomer joins a network through building connections and gains centrality. We provide three efficient heuristics that build edges between the newcomer and existing members of a social network and compare their performances in terms of three centrality metrics. We perform experiments on random graphs generated by two synthetic network models and four real-world networks. Not only our heuristics considerably outperform the random benchmark algorithm, but the results also reveal some new insights that are related to centrality and network topology. In particular, the results distinguish between measures along two dimensions: the first concerns with the number of centers in a network, and the second concerns with the type of involvement of a node in the network.

Information diffusion has been a crucial issue in social network research. The influence maximization problem asks for ways to identify nodes that are able to maximally diffuse information through a network. Recently, game-theoretical methods emerged that partitions a network into coalitions, and facilitate influence maximization. A notable example is a method based on Shapley value [13]. A major disadvantage with this approach is the lack of consideration of externality on coalition formation. In this paper, we propose consensus value for influence maximization (CVIM) to overcome this limitation. After forming coalitions using consensus value, the method selects initial coalitions based on the gain of each coalition, and finally chooses seed nodes in the selected coalitions. Empirical evidence over several real-world networks demonstrates that CVIM produces better results of influence maximization while significantly reducing the running time as compared to existing approaches.