Engineering Commons^™

Computational Model For Neural Architecture Search, Ram Deepak Gottapu

Doctoral Dissertations

"A long-standing goal in Deep Learning (DL) research is to design efficient architectures for a given dataset that are both accurate and computationally inexpensive. At present, designing deep learning architectures for a real-world application requires both human expertise and considerable effort as they are either handcrafted by careful experimentation or modified from a handful of existing models. This method is inefficient as the process of architecture design is highly time-consuming and computationally expensive.

The research presents an approach to automate the process of deep learning architecture design through a modeling procedure. In particular, it first introduces a framework that treats …

Applications Of Simulation And Optimization Techniques In Optimizing Room And Pillar Mining Systems, Angelina Konadu Anani

Doctoral Dissertations

"The goal of this research was to apply simulation and optimization techniques in solving mine design and production sequencing problems in room and pillar mines (R&P). The specific objectives were to: (1) apply Discrete Event Simulation (DES) to determine the optimal width of coal R&P panels under specific mining conditions; (2) investigate if the shuttle car fleet size used to mine a particular panel width is optimal in different segments of the panel; (3) test the hypothesis that binary integer linear programming (BILP) can be used to account for mining risk in R&P long range mine production sequencing; and (4) …

Approximation For Single-Channel Multi-Server Queues And Queuing Networks With Generally Distributed Inter-Arrival And Service Times, Carlos Roberto Chaves

Doctoral Dissertations

"This dissertation is divided into two papers. The first paper is related to developing a closed-form approximation for single-channel multiple-server queues with generally distributed inter-arrival and service times, which are often found in numerous settings, e.g., airports and manufacturing systems. Unfortunately, exact models for such systems require distributions for the underlying random variables. Further, data for fitting distributions is sometimes not available, and one only has access to means and variances of the underlying input random variables. Under heavy traffic, excellent approximations already exist for this purpose. In the first paper, a new approximation method for medium traffic is presented. …