Computational Engineering Commons^™

The Digital Neuron: Neural Cellular Automata For Neural–Symbolic Translation, Nicole Assenza

SMU Data Science Review

A neural cellular automata (NCA) architecture, referred to as Pluto’s NCA, was developed to characterize bilateral communication and semantic reciprocity between symbolic representations and a spatially distributed update field. The architecture employs an encoder–automata–decoder pipeline that maps symbolic inputs into a multichannel state field and reconstructs them through agreement-driven attractor convergence within a stable semantic attractor landscape. System behavior was evaluated under controlled perturbations, including rhythmic desynchronization, graded ablations, correlated and independent noise, and percolation-based structural degradation. Quantities such as Agreement(t), internal coherence Aᵢ(t), the recovery time constant τ, and the critical percolation threshold pc were measured to assess stability, …

Graph Neural Networks Powered Scientific Paper Recommendation, Junhao Shen

Computer Science and Engineering Theses and Dissertations

Scientific paper recommendation systems aim to help researchers discover relevant papers amidst the vast and ever-growing body of literature. With the exponential yearly increase in scientific publications, the demand for effective paper recommendation solutions has become both critical and increasingly challenging. In recent years, deep learning techniques have revolutionized recommender systems, and scientific paper recommendations have naturally integrated these advancements. In this dissertation, we address these challenges through three progressive contributions.

First, we enhance traditional content-based methods using Graph Neural Networks (GNNs) by introducing a Graph Convolutional Network-strengthened Topic Modeling (GCN-TM) approach. This method improves upon conventional topic modeling techniques …

Neuro-Symbolic Commonsense Reasoning With Resistance To Data Poisoning: A First-Order Logic And Sub-Symbolic Embeddings Framework, Bryce Shurts, King-Ip Lin

Computer Science and Engineering Theses and Dissertations

Commonsense reasoning has long presented a hurdle between conversational agents and their ability to naturally engage with humans in conversation, as the infinitely dimensional nature of a dialogue’s topics presents a significant reasoning challenge in the study of Natural Language Understanding (NLU). Such a system must conceivably be able to act as a generalizable system for evaluating and reasoning about commonsense statements, problems, and queries: in this way, the agent can attempt to quantify the reasonability of a given input. We attempt to address this through the integration of an explainable neuro-symbolic system that leverages Logical Tensor Networks (LTNs) and …

An Efficient Integrated Circuit Simulator And Time Domain Adjoint Sensitivity Analysis, Jiahua Li

Electrical Engineering Theses and Dissertations

In this paper, we revisit time-domain adjoint sensitivity with a circuit theoretic approach and an efficient solution is clearly stated in terms of device level. Key is the linearization of the energy storage elements (e.g., capacitance and inductance) and nonlinear memoryless elements (e.g., MOS, BJT DC characteristics) at each time step. Due to the finite precision of computation, numerical errors that accumulate across timesteps can arise in nonlinear elements.

Energy Dissipation In A Sand Damper Under Cyclic Loading, Ehab Sabi

Civil and Environmental Engineering Theses and Dissertations

Various seismic and wind engineering designs and retrofit strategies have been in development to meet structures' proper and safe operation during earthquake and wind excitation. One such method is the addition of fluid and particle dampers, such as sand dampers, in an effort to reduce excessive and dangerous displacements of structures. The present study implements the discrete element method (DEM) to assess the performance of a pressurized sand damper (PSD) and characterize the dissipated energy under cyclic loading. The idea of a PSD is to exploit the increase in shearing resistance of sand under external pressure and the associated ability …

Context-Aware Sensing And Fusion For Structural Health Monitoring And Night Time Traffic Surveillance, Xinxiang Zhang

Electrical Engineering Theses and Dissertations

Rapid developments in computer vision technologies have been transforming many traditional fields in engineering and science in the last few decades, especially in terms of diagnosing problems from visual images. Leveraging computer vision technologies to inspect, monitor, assess infrastructure conditions, and analyze traffic dynamics, has gained significant increase in both effectiveness and efficiency, compared to the cost of traditional instrumentation arrays to monitor, and manually inspect civil infrastructures and traffic conditions. Therefore, to construct the next-generation intelligent civil and transportation infrastructures, this dissertation develops a comprehensive computer-vision based sensing and fusion framework for structural health monitoring and intelligent transportation systems. …

Price Optimization For Revenue Maximization At Scale, Nikhil Gupta, Massimiliano Moro, Kailey A. Ayala, Bivin Sadler

SMU Data Science Review

This study presents a novel approach to price optimization in order to maximize revenue for the distribution market of non-perishable products. Data analysis techniques such as association mining, statistical modeling, machine learning, and an automated machine learning platform are used to forecast the demand for products considering the impact of pricing. The techniques used allow for accurate modeling of the customer’s buying patterns including cross effects such as cannibalization and the halo effect. This study uses data from 2013 to 2019 for Super Premium Whiskey from a large distributor of alcoholic beverages. The expected demand and the ideal pricing strategy …

Qlime-A Quadratic Local Interpretable Model-Agnostic Explanation Approach, Steven Bramhall, Hayley Horn, Michael Tieu, Nibhrat Lohia

SMU Data Science Review

In this paper, we introduce a proof of concept that addresses the assumption and limitation of linear local boundaries by Local Interpretable Model-Agnostic Explanations (LIME), a popular technique used to add interpretability and explainability to black box models. LIME is a versatile explainer capable of handling different types of data and models. At the local level, LIME creates a linear relationship for a given prediction through generated sample points to present feature importance. We redefine the linear relationships presented by LIME as quadratic relationships and expand its flexibility in non-linear cases and improve the accuracy of feature interpretations. We coin …

Generalized Relay Network Design And Collaborative Dispatching In Truckload Transportation, Amin Ziaeifar

Operations Research and Engineering Management Theses and Dissertations

The truckload industry faces a serious problem of high driver shortage and turnover rate which is typically around 100\%. Among the major causes of this problem are extended on-the-road times where drivers handle several truckload pickup and deliveries successively; non-regular schedules and get-home rates; and low utilization of drivers dedicated time. These are by-and-large consequences of the driver-to-load dispatching method, which is based on point-to-point dispatching or direct shipment from origin-to-destination, commonly employed in the industry. In this dissertation, we consider an alternative dispatching method that necessitates careful design of an underlying network. In this scheme, a truckload on its …

Improve Image Classification Using Data Augmentation And Neural Networks, Shanqing Gu, Manisha Pednekar, Robert Slater

SMU Data Science Review

In this paper, we present how to improve image classification by using data augmentation and convolutional neural networks. Model overfitting and poor performance are common problems in applying neural network techniques. Approaches to bring intra-class differences down and retain sensitivity to the inter-class variations are important to maximize model accuracy and minimize the loss function. With CIFAR-10 public image dataset, the effects of model overfitting were monitored within different model architectures in combination of data augmentation and hyper-parameter tuning. The model performance was evaluated with train and test accuracy and loss, characteristics derived from the confusion matrices, and visualizations of …