32584 research outputs found
Sort by
Spartan Daily, February 26, 2025
Get PDFVolume 164, Issue 15https://scholarworks.sjsu.edu/spartan_daily_2025/1014/thumbnail.jp
Unveiling the Transformative Power of Unsupervised machine learning through Clustering
Get PDFClustering methods demonstrated their transformative potential across various industries through image segmentation, anomaly detection, bioinformatics, and customer segmentation. The presentation explores these techniques in unsupervised machine learning, focusing on foundational clustering algorithms such as K-means, Hierarchical Clustering, and DBSCAN. Through an in-depth analysis of their underlying principles and computational intricacies, the presentation highlights how these methods have evolved to address complex, high-dimensional data problems. The presentation provides insights into how K-means remains a versatile tool for partitioning data in linear spaces. It delves into Hierarchical Clustering\u27s unique approach to building dendrograms and capturing multi-scale data relationships, and how DBSCAN\u27s density-based framework reveals clusters amidst noise, making it ideal for discovering patterns in irregular, real-world datasets. The presentation offers a comprehensive understanding of these algorithms and equips aspiring data scientists and industry professionals with the tools to harness the power of clustering for impactful, data-driven decisions.https://scholarworks.sjsu.edu/oer/1005/thumbnail.jp
Spartan Daily, May 6, 2025
Get PDFVolume 164, Issue 41https://scholarworks.sjsu.edu/spartan_daily_2025/1040/thumbnail.jp
Spartan Daily, May 8, 2025
Get PDFVolume 164, Issue 43https://scholarworks.sjsu.edu/spartan_daily_2025/1042/thumbnail.jp
Optimization of Permutation Flowshop Scheduling Using an Island Genetic Algorithm for Makespan Minimization
No full textThe Permutation Flowshop Scheduling Problem is a well-known NP-hard combinatorial optimization problem that involves the sequencing of n jobs across m machines in the same order to minimize the total makespan value. This project proposes a Heterogeneous Island Genetic Algorithm framework (HIGA). Each island represents a group of solutions that evolve in parallel using different initialization heuristics, crossover and mutation operators, and adaptive parameters. A dynamic, stagnation-based migration strategy is proposed to maintain targeted communication between the islands. The proposed HIGA approach was compared against the basic Standard Genetic Algorithm (SGA) and a more advanced Niche-based Genetic Algorithm (NEH-NGA) on Taillard’s benchmark dataset. Experimental results indicate HIGA effectively balances solution quality and efficiency, matching the best-known makespan value or coming close to it, particularly for larger instances, while being several-fold faster than NEH-NGA and achieving significantly better results than the SG
Advanced Knowledge Extraction with Biomedical Data Using LLMs
No full textThe rapid growth of biomedical research has led to an overwhelming volume of unstructured textual data in the scientific literature. This has necessitated the development of an automated approach for knowledge extraction and integration. In
this project, we present a comprehensive pipeline for constructing a unified biomed- ical knowledge graph by combining two well-known datasets: CHEMPROT [1],
which captures chemical–protein interactions, and EU-ADR [2], which annotates drug–gene–disease relationships. In order to identify important biomedical entities and interactions from CHEMPROT dataset, we perform Named Entity Recognition (NER) and relation Extraction (RE) using state-of-the-art biomedical models like BioBERT [3], BioGPT [4] and bert-base-cased [5] . NER step identifies 647 unique chemicals and 790 gene/protein mentions, while the RE step produces 10,558 relation triples. We create a heterogeneous knowledge graph linking chemical, genes/proteins and disease. The final graph created consisted of 3194 nodes and 7716 edges. Followed by applying multi-hop reasoning(3-hop, 4-hop) to infer new knowledge and relationships, helping to uncover new biomedical insights that are not explicitly mentioned in the literature [6]. Our results show how multi-hop inference enhances the knowledge graph beyond direct annotations
Unconditional-to-conditional Transfer and Optimization for Web-based Skybox GAN
No full textGenerative adversarial networks (GANs) are known for their ability to generate high qualityimages mimicking real life or even particular art styles. Yet for all their capability, casuallytraining a GAN on an average machine can be infeasible as GANs require an enormous amountof time and data to train. Even with a trained GAN, model inference demands heavycomputations, making GANs difficult to deploy on applications. To address these limitations,techniques such as transfer learning and quantization have been leveraged to speed up training ofGANs and lighten computational cost of GAN inference. This project aims to use suchtechniques to efficiently train and optimize a sky image GAN for deployment on a skyboxgenerator web page. We conducted experiments which demonstrated that transfer learning, in theform of direct weight splicing from a pre-trained unconditional model to a conditional model,accelerates the conditional model’s training. After 100 thousands of images (kimg) of training,the model with transferred weights achieved an FID score of 29.01, outperforming theconditional model trained from scratch which obtained an FID score of 134.51. As forquantization, our results appear less impressive as the GAN model size of 120 megabytes isshrunk only to 117 megabytes, and inference speed seems to remain unaffected. The final webpage deploys this model through ONNX Runtime Web and presents an interface allowing usersto generate skyboxes based on cloud types, fulfilling a use case of a choice-based AI skyboxgenerator
Telling the Library’s Story: A Step-by-Step Tutorial for Using Data Visualization to Show Impact on Teaching and Learning
Get PDFThis tutorial introduces school librarians to the use of data visualization tools for documenting and communicating their impact on teaching and learning. By leveraging accessible platforms like Google Forms and Google Sheets, the authors demonstrate how librarians can build dashboards and real-time visual reports to showcase co-teaching, instructional collaboration, and student engagement. The tutorial features real-world examples, highlights common data sources already available to librarians, and argues that dynamic visual storytelling is a powerful alternative to traditional library statistics
The Conventional Emptiness of Human Beings: Reviewing Archer’s Theory of Agency through Nagarjuna’s Looking Glass
Get PDFMargaret Archer is probably the critical realist who has most significantly affected the debate around the relationship between social structure and agency. In the volume, Being Human: The Problem of Agency, she aims to better define what is meant by the properties and causal powers of agency itself, i.e., self-consciousness and reflexivity. In this paper, this theoretical attempt is reviewed through the critical standpoint of Nagarjuna. Nagarjuna’s philosophy appears to be a particularly compelling tool for criticizing Archer’s theory. Claiming the emptiness of everything, the founder of the Buddhist Madhyamaka school provides a critical perspective for revising the distinctiveness of human properties and causal powers. Specifically, Nagarjuna’s rejection of the intrinsic nature of all entities reveals that Archer’s attempt to discern something as inherently human is questionable. If our sense of self emerges from embodied practice in the natural world, it may be that it cannot be demolished by social and linguistic forces. However, this does not mean that it can be regarded as a sui generis human property and power. That same embodied practice in the world makes human beings empty of inherent nature. Nevertheless, since emptiness itself is empty, we cannot conclude that human beings are inherently empty, but merely conventionally empty
Spartan Daily, May 1, 2025
Get PDFVolume 164, Issue 40https://scholarworks.sjsu.edu/spartan_daily_2025/1039/thumbnail.jp