No abstract

The dream of building intelligent robotic systems to interact and communicate with people and help them in their lives is a very old and ongoing study. In this research, the massive parallel autonomous checkers agent "MPACA" can autonomously play checkers with a human upto Grandmaster level without requiring a special checkers board for detecting human movements. The main aim and contribution of this research is proposing enhanced algorithms for a game tree search using two different approaches. The first was a task-based approach on CPU with a parallel database, while the second was a threads-based approach on the GPU with no divergence and dynamic parallelism. The two approaches were compared with previous studies using various approaches, including threads on CPU for up to 6x speedup for an 8-core processor and threads on GPU using iterative dependence and fixed grid and block size of up to 40x speedup at 14 depth. Furthermore, the approaches were tested with different depths on the CPU and the GPU. The result shows speed up for parallel CPU tasks up to 7x for an 8core processor and parallel GPU of up to 80x at 14 depth.

Abstract-In this paper, a comprehensive survey on gaming tree searching methods that can use to find the best move in two players zero-sum computer games was introduced. The purpose of this paper is to discuss, compares and analyzes various sequential and parallel algorithms of gaming tree, including some enhancement for them. Furthermore, a number of open research areas and suggestions of future work in this field are mentioned.

Parallel performance for GPUs today surpasses the traditional multi-core CPUs. Currently, many researchers started to test several AI algorithms on GPUs instead of CPUs, especially after the release of libraries such as CUDA and OpenCL that allows the implementation of general algorithms on the GPU. One of the most famous game tree search algorithms is Negamax, which tries to find the optimal next move for zero sum games. In this research, an implementation of an enhanced parallel NegaMax algorithm is presented, that runs on GPU using CUDA library. The enhanced algorithms use techniques such as no divergence, dynamic parallelism and shared GPU table. The approach was tested in checkers and chess games. It was compared with previous studies, including threads on CPU for up to 6x speedup for an 8 core processor and threads on GPU using iterative dependence and fixed grid and block size for up to 40x speedup at 14 depth. Furthermore, the approach was tested with different depths on the CPU and the GPU. The result shows speed up for parallel GPU up to 80x at 14 depth for checkers game and 90x at 14 depth for chess game, which doubled the previous research results.

Successful Prediction for MHC Class II epitopes is an essential step in designing Genetic Vaccines[1]. MHC Class II epitopes are short peptides with length between 9 and 25 amino acids which are bound by MHC. These epitopes are recognized by T-Cell Receptors and leads to activation of cellular and humoral immune system and, ultimately, to effective destruction of pathogenic organism. Successful prediction of MHC class II epitopes is more difficult than MHC class I epitopes due to open binding groove at both ends in class II molecules, this structure leads to variable length for MHC II epitopes and complicating the task for detecting the core binding 9-mer. Large efforts have been exerted in developing algorithms to predict which peptides will bind to a given MHC class II molecules. In this paper we presented a novel classification algorithm for predicting MHC Class II epitopes using Multiple Instance Learning technique. Separated Constructive Clustering Ensemble (SCCE) is our new version for Constructive Clustering Ensemble (CCE)[27]. This method was used for converting multiple instance learning problem into normal Single Instance Problem. Most of the processing in this method lies mainly in vector preparation step before using classifier; Support Vector Machine (SVM) has been used as a method with proven performance in a wide range of practical applications[38]. SCCE integrated many algorithms like Genetic Algorithm, K medoid clustering, Ensemble learning and Support vector machine in an orchestration to predict the MHC II epitopes. SCCE was tested over three benchmark data sets and proved to be very competitive with the state of art regression methods. SCCE achieved these results using only binder and non binder flags; without need for regression data. An implementation of MHCII-SCCE as an online web server for predicting MHC-II Epitopes is freely accessible at http://www.fci.cu.edu.eg:8080/MHCII_Server/MHCII SCCE Server 1.0.htm.