Showing 1 - 10 of 13 Items

Date: 2022-01-01

Creator: Daniel Little

Access: Open access

In the recent past, extensive experimental works have been performed to predict joint voting outcomes in Congress based on a game-theoretic model of voting behavior known as Linear Influence Games. In this thesis, we improve the model selection and evaluation procedure of these past experiments. First, we implement two methods, Nested Cross-Validation with Tuning (Nested CVT) and Bootstrap Bias Corrected Cross-Validation (BBC-CV), to perform model selection and evaluation with less bias than previous methods. While Nested CVT is a commonly used method, it requires learning a large number of models; BBC-CV is a more recent method boasting less computational cost. Using Nested CVT and BBC-CV we perform not only model selection but also model evaluation, whereas the past work was focused on model selection alone. Second, previously models were hand picked based on performance measures gathered from CVT, but both Nested CVT and BBC-CV necessitate an automated model selection procedure. We implement such a procedure and compare its selections to what we otherwise would have hand picked. Additionally, we use sponsorship and cosponsorship data to improve the method for estimating unknown polarity values of bills. Previously, only subject code data was used. This estimation must be done when making voting outcome predictions for a new bill as well as measuring validation or testing errors. We compare and contrast several new methods for estimating unknown bill polarities.

Date: 2017-05-01

Creator: Dominick Sanchez

Access: Open access

Particle Swarm Optimization (PSO) is often used for optimization problems due to its speed and relative simplicity. Unfortunately, like many optimization algorithms, PSO may potentially converge too early on local optima. Using multiple neighborhoods alleviates this problem to a certain extent, although premature convergence is still a concern. Using dynamic topologies, as opposed to static neighborhoods, can encourage exploration of the search space at the cost of exploitation. We propose a new version of PSO, Dynamic-Static PSO (DS-PSO) that assigns multiple neighborhoods to each particle. By using both dynamic and static topologies, DS-PSO encourages exploration, while also exploiting existing knowledge about the search space. While DS-PSO does not outperform other PSO variants on all benchmark functions we tested, its performance on several functions is substantially better than other variants.

Date: 2018-05-01

Creator: Corinne Alini

Access: Open access

Computer Vision involves the extraction of data from images that are analyzed in order to provide information crucial to many modern technologies. Object recognition has proven to be a difficult task and programming reliable object recognition remains elusive. Image processing is computationally intensive and this issue is amplified on mobile platforms with processor restrictions. The real-time constraints demanded by robotic soccer in RoboCup competition serve as an ideal format to test programming that seeks to overcome these challenges. This paper presents a method for ball recognition by analyzing the movement of the ball. Major findings include enhanced ball discrimination by replacing the analysis of static images with absolute change in brightness in conjunction with the classification of apparent motion change.

Date: 2019-05-01

Creator: James Little

Access: Open access

Isolation-Based Scene Generation (IBSG) is a process for creating synthetic datasets made to train machine learning detectors and classifiers. In this project, we formalize the IBSG process and describe the scenarios—object detection and object classification given audio or image input—in which it can be useful. We then look at the Stanford Street View House Number (SVHN) dataset and build several different IBSG training datasets based on existing SVHN data. We try to improve the compositing algorithm used to build the IBSG dataset so that models trained with synthetic data perform as well as models trained with the original SVHN training dataset. We find that the SVHN datasets that perform best are composited from isolations extracted from existing training data, leading us to suggest that IBSG be used in situations where a researcher wants to train a model with only a small amount of real, unlabeled training data.

Date: 2024-01-01

Creator: Cassandra Goldberg

Access: Open access

This project explores novel approaches for Synthetic Aperture Radar (SAR) image segmentation that integrate established statistical properties of SAR into deep learning models. First, Perlin Noise and Generalized Gamma distribution sampling methods were utilized to generate a synthetic dataset that effectively captures the statistical attributes of SAR data. Subsequently, deep learning segmentation architectures were developed that utilize average pooling and 1x1 convolutions to perform statistical moment computations. Finally, supervised and unsupervised disparity-based losses were incorporated into model training. The experimental outcomes yielded promising results: the synthetic dataset effectively trained deep learning models for real SAR data segmentation, the statistically-informed architectures demonstrated comparable or superior performance to benchmark models, and the unsupervised disparity-based loss facilitated the delineation of regions within the SAR data. These findings indicate that employing statistically-informed deep learning methodologies could enhance SAR image analysis, with broader implications for various remote sensing applications and the general field of computer vision. The code developed for this project can be found here: https://github.com/cgoldber/Statistically-Principled-SAR-Segmentation.git.

• Embargo End Date: 2025-05-16

Date: 2024-01-01

Creator: Darien Gillespie

Access: Embargoed

Date: 2022-01-01

Creator: Stephen Crawford

Access: Open access

In the past decade, numerous datasets have been released with the explicit goal of furthering non-intrusive load monitoring research (NILM). NILM is an energy measurement strategy that seeks to disaggregate building-scale loads. Disaggregation attempts to turn the energy consumption of a building into its constituent appliances. NILM algorithms require representative real-world measurements which has led institutions to publish and share their own datasets. NILM algorithms are designed, trained, and tested using the data presented in a small number of these NILM datasets. Many of the datasets contain arbitrarily selected devices. Likewise, the datasets themselves report aggregate load information from building(s) which are similarly selected arbitrarily. This raises the question of the representativeness of the datasets themselves as well as the algorithms based on their reports. One way to judge the representativeness of NILM datasets is to look for the presence of outliers in these datasets. This paper presents a novel method of identifying outlier devices from NILM datasets. With this identification process, it becomes possible to mitigate and measure the impact of outliers. This represents an important consideration to the long-term deployment of NILM algorithms.

Date: 2017-05-01

Creator: William Gantt

Access: Access restricted to the Bowdoin Community

Date: 2019-05-01

Creator: Kevin Fakai Chen

Access: Open access

Particle Swarm Optimization (PSO) is a widely-used nature-inspired optimization technique in which a swarm of virtual particles work together with limited communication to find a global minimum or optimum. PSO has has been successfully applied to a wide variety of practical problems, such as optimization in engineering fields, hybridization with other nature-inspired algorithms, or even general optimization problems. However, PSO suffers from a phenomenon known as premature convergence, in which the algorithm's particles all converge on a local optimum instead of the global optimum, and cannot improve their solution any further. We seek to improve upon the standard Particle Swarm PSO algorithm by fixing this premature convergence behavior. We do so by storing and exploiting increased information in the form of past bests, which we deem enhanced memory. We introduce three types of modifications to each new algorithm (which we call a GEM-PSO: Particle Swarm Optimization, Guided by Enhanced Memory, because our modifications all deal with enhancing the memory of each particle). These are procedures for saving a found best, for removing a best from memory when a new one is to be added, and for selecting one (or more) bests to be used from those saved in memory. By using different combinations of these modifications, we can create many different variants of GEM-PSO that have a wide variety of behaviors and qualities. We analyze the performance of GEM-PSO, discuss the impact of PSO's parameters on the algorithms' performances, isolate different modifications in order to closely study their impact on the performance of any given GEM-PSO variant, and finally look at how multiple modifications perform. Finally, we draw conclusions about the efficacy and potential of GEM-PSO variants, and provide ideas for further exploration in this area of study. Many GEM-PSO variants are able to consistently outperform standard PSO on specific functions, and GEM-PSO variants can be shown to be promising, with both general and specific use cases.

Date: 2020-01-01

Creator: Dylan Hayton-Ruffner

Access: Open access

During the course of research, scholars often explore large textual databases for segments of text relevant to their conceptual analyses. This study proposes, develops and evaluates two algorithms for automated concept detection in theoretical corpora: ACS and WMD retrieval. Both novel algorithms are compared to key word retrieval, using a test set from the Digital Ricoeur corpus tagged by scholarly experts. WMD retrieval outperforms key word search on the concept detection task. Thus, WMD retrieval is a promising tool for concept detection and information retrieval systems focused on theoretical corpora.