41530 research outputs found
Sort by
Pain Points: Cluster Analysis In Chronic Pain Networks
Get PDFChronic pain is a pervasive health issue, affecting a significant portion of the population and posing complex challenges due to its diverse etiology and individualized impact. To address this complexity, there is a growing interest in grouping chronic pain patients based on their unique treatment needs. While various methodologies for patient grouping have emerged, leveraging graph-based approaches to produce and evaluate such groupings remains largely unexplored. Recent studies have shown promise in integrating knowledge graphs into exploring patient similarity across different biological domains, indicating potential avenues for research. Additionally, there is a growing interest in investigating patient similarity networks, highlighting the importance of innovative approaches to understanding chronic pain.
Graphs offer a transparent and easily interpretable framework for analyzing patient classifications, providing valuable insights into underlying patterns and connections. By leveraging graph theory, this thesis proposes a novel approach to address the terminological disparities that exist across disciplines studying chronic pain. By constructing a graph of pain-related terminology sourced from interdisciplinary literature, we aim to facilitate link prediction and clarify connections among disparate terminologies. This approach seeks to bridge disciplinary divides, fostering a cohesive understanding of chronic pain and promoting collaborative efforts toward effective management and treatment strategies.
Through the integration of graph theory and interdisciplinary research, this thesis contributes to advancing our understanding of chronic pain and lays the groundwork for future explorations in patient grouping and treatment optimization by proposing a graph-based clustering method as well as a method for evaluating the robustness of a cluster
Contrastive Filtering And Dual-Objective Supervised Learning For Novel Class Discovery In Document-Level Relation Extraction
Get PDFRelation extraction (RE) is a task within natural language processing focused on the classification of relationships between entities in a given text. Primary applications of RE can be seen in various contexts such as knowledge graph construction and question answering systems. Traditional approaches to RE tend towards the prediction of relationships between exactly two entity mentions in small text snippets. However, with the introduction of datasets such as DocRED, research in this niche has progressed into examining RE at the document-level. Document-level relation extraction (DocRE) disrupts conventional approaches as it inherently introduces the possibility of multiple mentions of each unique entity throughout the document along with a significantly higher probability of multiple relationships between entity pairs.
There have been many effective approaches to document-level RE in recent years utilizing various architectures, such as transformers and graph neural networks. However, all of these approaches focus on the classification of a fixed number of known relationships. As a result of the large quantity of possible unique relationships in a given corpus, it is unlikely that all interesting and valuable relationship types are labeled before hand. Furthermore, traditional naive approaches to clustering on unlabeled data to discover novel classes are not effective as a result of the unique problem of large true negative presence. Therefore, in this work we propose a multi-step filter and train approach leveraging the notion of contrastive representation learning to discover novel relationships at the document level. Additionally, we propose the use of an alternative pretrained encoder in an existing DocRE solution architecture to improve F1 performance in base multi-label classification on the DocRED dataset by 0.46.
To the best of our knowledge, this is the first exploration of novel class discovery applied to the document-level RE task. Based upon our holdout evaluation method, we increase novel class instance representation in the clustering solution by 5.5 times compared to the naive approach and increase the purity of novel class clusters by nearly 4 times. We then further enable the retrieval of both novel and known classes at test time provided human labeling of cluster propositions achieving a macro F1 score of 0.292 for novel classes. Finally, we note only a slight macro F1 decrease on previously known classes from 0.402 with fully supervised training to 0.391 with our novel class discovery training approach
Analog Speech Recognition Circuit
Get PDFThis project performs speech recognition using analog circuits to compute the Fourier transform of a voice signal. There is a digital portion of the design to perform a comparison between the desired Fourier transforms and the voice signal’s Fourier transform. Due to time limitations, the project is limited to a set command that the circuit can execute. The design is also focused on the vocal frequencies of women due to the stark difference between men and women’s voices. Analog circuits are not being used commercially for speech recognition at this time, but there is research being done on methods to do so. The final design achieves an accuracy of 54.57% when tested on women saying the word “on”. The system was built around the frequencies found in audio recordings of the word “on”, and because of that it will only function consistently when the word is intentionally said and each syllable is clearly pronounced. Because the system relies on comparators to determine whether the word is present, the word must also be said loudly enough due to the gain limitations of the microphone setup used
A Comparative Study Of The NPM, PyPI, Maven, And RubyGems Open-Source Communities
Get PDFOpen-source software (OSS) ecosystems, defined as environments composed of package managers and programming languages (e.g., NPM for JavaScript), are essential for software development and foster collaboration and innovation. Although their significance is acknowledged, understanding what makes OSS communities healthy and sustainable requires further exploration. This thesis quantitatively assesses the health of OSS projects and communities within the NPM, PyPI, Maven, and RubyGems ecosystems. We explore five research questions addressing project standards, community responsiveness, contribution distribution, contributor retention, and newcomer integration strategies. Our analysis shows varied documentation practices, insider engagement levels, and contribution patterns. Our findings highlight both strengths and different areas for improvement across ecosystems. For example, RubyGems excels in the adoption of project documentation and exhibits the most even distribution of contributions among all contributors, including highly active contributors. and a very responsive community, but it needs to improve contribution retention and attract newcomers to the projects. Meanwhile, NPM and Maven show a trend toward getting new contributors, characterized by a high ratio of individual contributions. They need to better adopt a code of conduct, pull request templates, and increase the number of active contributors in a project. This thesis offers insights to developers and maintainers on how to strengthen ecosystems and support vibrant communities effectivel