فهرست مطالب بهنام قوامی

The modern processors that consist of large caches are very vulnerable to transient errors. Due to the importance of this problem, coding methods are adopted for protection against errors. The cost of reliability should be incurred to optimize the use of energy and area. This paper proposes an approach to dependability enhancement in unprotected caches of processors. For this purpose, the low-cost tag error mitigation mechanism is adopted to analyze the reliability of caches in modern processors. Benefiting from the tag Hamming distance increase technique, the proposed approach has a much lower overhead and decreases the false hit rate to zero.

Recently, a number of Extreme Learning Machine (ELM) based training algorithms have been introduced for training deep neural network structures. ELM based Auto-Encoder (ELM-AE) is one such algorithm that has been used for making multilayer structures and tuning parameters of each layer. In a simple ELM-AE training algorithm, the weights of the first layer are initialized randomly. This issue is a leading factor in producing reconstruction error. The frequent use of ELM-AE in deep network layers results in propagating such errors through deep structures and in decreasing performance as a consequent. In this paper, we introduce a multilayer structure and a new learning algorithm to train it that prevents error propagation. In order to boost the performance of the model, the parameters in the first layer are initialized by a novel type of ELM-AE called Repeated-AE (RAE) rather than by a random selection method. This RAE-based technique determines the parameters in the first layer far better than do the other ELM-AE existed methods. Next, a single hidden layer ELM is applied for handling the classification task. Experimental results for data classification show that the proposed method outperforms some other methods in terms of the average accuracy over all datasets by amounts of 4%, 26%, 17% and 31%. Eventually, so as to verify the performance of the proposed multilayer ELM-AE in application, we used this model to reconstruct images. The reconstructed images obtained by our approach appeared visually a lot better compared to those obtained by the other methods do.

The inference of Gene Regulatory Network (GRN) using gene expression data is significantly important in order to understand gene dependencies, regulatory functions among genes, biological processes, way of process occurrence and avoiding some unplanned processes (disease). The accurate inference of GRN needs the accurate inference of predictor set. Generally, the main limitations of the predictor set inference are the small number of samples, the large number of genes and also the possibility influence of noise in gene expression data. Hence, providing efficient methods to infer predictor set with high reliability is a serious need. In this paper, an efficient method is proposed to infer predictor set using Gravitational Search Algorithm (GSA). A GSA is used for each target gene to infer the predictor subset of the gene. In a population, a mass represents a predictor subset of the associated gene. The initial population per target gene is generated by Pearson Correlation Coefficient (PCC). In order to guide the GSA, Mean Conditional Entropy (MCE) is used as the assessment criterion. Experimental results show that the proposed method has a good ability to infer the predictor set with high reliability. In addition, we also compared the proposed algorithm with a recent similar method based on genetic algorithm. Comparison results reveal the advantage of the proposed algorithm on biological datasets with small data volumes and large network scales.