Sharif Digital Repository

Nowadays, thanks to improvement of processing hardware and plenty of data available, artificial intelligence and specifically Deep Learning are being one of the powerful tools for solving different problems. Also graph is one of the powerful tools for modeling different data structures. Graph matching is one of the problems in the field of graph problems.In this thesis we consider the problem of graph matching in Erdos-Renyi graphs. The graph matching problem refers to recovering the node-to-node correspondence between two correlated graphs. Previous works theoretically showed that recovering is feasible in sparse Erdos-R´enyi graphs if and only if the probability of having an edge between a... 

The main goal in many fields in the empirical sciences is to discover causal relationships among a set of variables from observational data. PC algorithm is one of the promising solutions to learn underlying causal structure by performing a number of conditional independence tests. In this paper, we propose a novel GPU-based parallel algorithm, called cuPC, to execute an order-independent version of PC. The proposed solution has two variants, cuPC-E and cuPC-S, which parallelize PC in two different ways for multivariate normal distribution. Experimental results show the scalability of the proposed algorithms with respect to the number of variables, the number of samples, and different graph... 

The loss function form in deep networks has been the subject of numerous studies in machine learning communities in recent years. Despite the fact that the problem of optimization in these networks has high dimensions and usually non-convex, gradient-based algorithms have acceptable results in practical experiments. In other words, the convergence point is usually a low-cost local minimum, and conventional optimization algorithms do not get stuck in high-cost local minima. It has been shown that this phenomenon occurs if the number of network weights is much greater than the number of training samples, and it is referred to as the over parameterization region. Recently, several studies have... 

One of the key objectives in many fields in machine learning is to discover causal relationships among a set of variables from observational data. In linear non-Gaussian acyclic models (LiNGAM), it can be shown that the true underlying causal structure can be identified uniquely from merely observational data. DirectLiNGAM algorithm is a well-known solution to learn the true causal structure in high dimensional setting. DirectLiNGAM algorithm executes in a sequence of iterations and it performs a set of comparisons between pairs of variables in each iteration. Unfortunately, the runtime of this algorithm grows significantly as the number of variables increases.In this thesis, we propose a... 

Nowadays we are witnessing the emergence of machine learning in various applications. One of the key problems in data science and machine learning is the problem of dimensionality reduction, which deals with finding a mapping that embeds samples to a lower-dimensional space such that, the relationships between the samples and their properties are preserved in the secondary space as much as possible. Obtaining such mapping is essential in today's high-dimensional settings. Moreover, due to the large volume of data and high-dimensional samples, it is infeasible or insecure to process and store all data in a single machine. As a result, we need to process data in a distributed manner.In this... 

There has been a sudden emergence of next generation networks in the past decade where the primary purposes are data ggregation/mining, distributed information and signal processing, and environmental control and monitoring. The distributed algorithms operating in such networks, should have simple structure and be robust against node failures or network dynamics.Extensive studies on designing and analyzing these algorithms have resulted in introducing different models of distributed systems with similar properties such as gossip algorithms, population protocols, and cellular automata-based systems. In this dissertation, we take first steps toward understanding the computational power of... 

Cable-driven robots are referred to as parallel robots actuated with cables. In fact,in such robots rigid links are replaced by cables, which may be extended to desired lengths without making mechanism much heavy. This robot possesses a number of unique properties that makes it suitable for many industrial applications. The main factor which makes cable robots analysis different from other parallel robots is the incapability of cables to push objects. Hence, obtaining the workspace of a cable robot is one of the most important subjects associated with this type of robot.The goal of this thesis is to design a spatial translational cable driven robot, which may be used for object handling. For... 

Uncertain fuel prices and also global climate changes are accompanied by some state initiatives such as renewable portfolio standards (RPS). This has caused a fast growth in the amount of renewable energy installed worldwide especially wind energy over last decades. However, the intrinsic characteristics of wind farms output power, i.e. intermittency and volatility of wind speed along with being non-dispatchable in output generation of wind turbine raises many new technical and financial challenges for power system operators and planners. Up to these issues, large level of wind power penetration causes a growing concern in wind energy curtailment issue which the wind farm’s operator may be...