32584 research outputs found
Sort by
Spartan Daily, April 30, 2025
Get PDFVolume 164, Issue 36https://scholarworks.sjsu.edu/spartan_daily_2025/1038/thumbnail.jp
The Effects of the COVID-19 Pandemic on Transit in the San Francisco Bay Area
Get PDFThis report presents the findings from our study for the California State Assembly Transportation Committee on the effects of the COVID-19 pandemic on Bay Area Transit. The study consisted of a review of the literature on the effects of the pandemic on transit in the US, a detailed look at changes in ridership and economics for Bay Area transit agencies, comparisons of Bay Area transit ridership changes to those in similar US metropolitan areas, and analysis of possible remedies to restore the financial health of Bay Area transit agencies. Bay Area transit ridership has recovered somewhat from the depths of the pandemic, but remains below pre-pandemic levels: bus ridership is currently 20 percent lower and rail ridership is 60 percent lower than before 2020. Much of the pre-pandemic transit ridership, especially on rail, was from professional and technical workers; many of these have continued to work at home, even after the official end of the pandemic. Federal funding provided short-term operating funding relief transit, but several agencies—BART, SFMTA, and Caltrain in particular—face severe funding shortfalls beginning in fiscal year 2026. Part of the shortfalls could be made up through a combination of fare increases and service reductions, but high fixed operating costs make it impossible for these agencies to remain viable without additional outside funding. The feasible funding sources from a financial standpoint are (1) a surcharge to bridge tolls, (2) an additional sales tax, or (3) a combination of toll surcharges and sales taxes
Artificial Intelligence in Modern Game Development
Get PDFArtificial intelligence has transformed the landscape of game development in ways early pioneers could only imagine. In my project I explore four major phases of AI evolution: classical AI methods, the rise of machine learning, the advent of transformer architectures, and a spirited debate about whether transformers truly think. To illustrate these shifts I use examples like MYCIN, which helped doctors decide treatments, and Deep Blue, which beat a world chess champion. You’ve probably lost track of time exploring an open world that adapts to your choices, right? That sense of immersion is powered by AI behind the scenes. From procedurally generated dungeons in indie titles to branching narratives in blockbuster games, AI shapes how much agency you feel. Have you ever wondered how matchmaking algorithms decide which opponents you face online, or how toxic behavior gets spotted? Live-service games rely on machine learning to keep communities fun and fair. The broad impact of AI in gaming can be seen in how modern open world titles adapt world conditions based on player choices. From dynamic weather systems to realistic economies AI has shaped player agency and immersion
The Cost of Creative Freedom: Comparing AAA and Indie Game Development
Get PDFThe independent development model provides developers with creative freedom and personal fulfillment because it enables them to learn through hands-on experience while creating distinctive experiences and developing diverse skills. It provides them with both long-term ownership of their work and the chance to achieve remarkable success through their own efforts while allowing them to maintain direct communication with players who appreciate their creative output. The choice between AAA and indie depends on individual career goals, but developers who view games as more than just products of entertainment can find greater meaning in the indie path, Long-term triple-A developers who want to find more fulfillment in their work often transition to indie because this path allows them to tell stories and innovate while maintaining ownership of their work. In conclusion, while AAA studios create the biggest pieces of entertainment for wide audiences that are highly advertised and talked about, indie developers maintain the most freedom to express creativity, which turns into the most beautiful, innovative. and engaging game worlds that high production could not replicat
A Comprehensive Study of Impacts of “Q” Bus Rapid Transit System on Blackstone Avenue
Get PDFThis study examines the dual impact of the Fresno Area Express (FAX) Q Line, a 15.7-mile Bus Rapid Transit (BRT) system launched in 2018, on the Fresno housing market and passenger satisfaction. The Q Line, designed to modernize public transit, features eco-friendly compressed natural gas (CNG) vehicles, real-time passenger information, and improved service efficiency. Housing market analysis focused on residential properties sold between 2012 and 2024, utilizing Geographic Information System(GIS) mapping to segment properties into three regions: the Q Line corridor, an outer buffer zone, and the rest of Fresno. Results indicate no statistically significant increase in property values within the Q Line corridor, challenging assumptions that public transit improvements substantially influence housing prices. Passenger satisfaction was assessed using FAX survey data and advanced machine learning models to identify key predictors. Statistical independence tests revealed no significant differences in satisfaction based on age, gender, employment status, or education. However, larger households reported higher satisfaction, while lower-income passengers expressed dissatisfaction. Machine learning models highlighted eight key factors influencing satisfaction, including audio-visual quality, value, comfort, closeness to home, and driver helpfulness. The findings underscore the importance of inclusive, passenger-centered transit services to enhance satisfaction while addressing disparities among lower-income riders. Policy recommendations encourage continued investment in public transit improvements without expecting significant housing market impacts. By prioritizing the identified satisfaction drivers, FAX can further improve the Q Line’s service quality and better meet the community’s needs, ensuring its long-term success
Augmenting Missing Sensor Data for Robust Human Activity Recognition
No full textApplications of ubiquitous computing, including health monitoring, sports analytics, and ambient-assisted living, rely on Human Activity Recognition (HAR) using wearable sensors. However, model robustness is challenged by missing sensor values, class imbalance, inter-subject variability, and temporal noise. This work proposes a complete HAR pipeline that addresses these challenges through sampling, time-series augmentation, dynamic feature handling, and GAN-PCA-based imputation. Built on the DeepSense architecture, the model integrates convolutional feature extraction with bi-GRUs for temporal modeling. The system is evaluated using 5-fold cross-validation, subject-aware holdout, and LOSEO strategies on the Opportunity dataset. Results demonstrate consistent accuracy across folds and strong generalization to unseen sessions and subjects, significantly outperforming baseline models. This study highlights the effectiveness of hybrid deep learning approaches in real-world HAR scenario
Malware Opcode Embedding and Quality Assessment of Generative Sample Embeddings
No full textMalware is software used to damage and disrupt computer systems with the intent to cause damage to the victim. Malware detection and classification into malware families is a crucial problem for cybersecurity researchers. One of the major bottlenecks in improving these systems is the shortage of good quality labeled malware data, especially for malware families with scarce samples. Researchers have utilized generative models to generate malware data to address this issue. Malware embeddings encode patterns within a malware file, which can be used to detect and classify malware. Recently, encouraging results have been obtained in generating malware embeddings using generative models. The experiments presented in this report aim to create high-quality malware opcode embeddings and then perform robust evaluations to assess their quality. The project seeks to generate high-quality malware embeddings that could be utilized to train malware detection and classification models
Comparative Analysis of Adversarial Permeability in CPU-native, QAT and ONNX-based Quantized Transformer Models
No full textThis work explores securing and optimization of Transformer-based time series forecasting models. We employ several quantization techniques, including quantization-aware training (QAT), and tested the robustness of quantized models by adversarially attacking them. The preliminary results of this work, in our controlled setup, indicate that quantized models outperform the full precision model in terms of robustness against adversarial attacks. They achieved this robustness while showing a very minimal decrease in their forecasting performance
DISEASE DIAGNOSIS USING RAG LLM WITH SMART PROMPT ENGINEERING
No full textAlthough recent trends indicate that LLMs outperform traditional methods in solving complex problems with enhanced reasoning, there has been barely any progress in replicating the quality of diagnoses like those of actual human doctors. The identification of an accurate diagnosis with thorough reasoning is still a significant challenge, even with advanced AI models. The process of performing accurate diagnosis remains challenging due to a lack of transparency in state-of-the-art models existing today, a lack of explanation in the diagnosis process, an emphasis on results rather than reasoning, and a lack of foundational knowledge in models, along with limited exploration of diseases. To solve these problems, we propose a RAG model with smart prompt engineering to develop a sound medical diagnostic agent. The rapidly evolving techniques in LLMs, particularly RAG models, have shown promising results in processing and interpreting complex data. With the use of smart prompt engineering techniques and the use of a RAG framework with strong databases, we have achieved desirable results and enhanced the diagnostic reasoning performance. Deepseek-R1-Distill-Qwen-7B, Mixtral-8x7B, and MedAlpaca were trained as RAG models with PubMed articles and PMC patient data by varying the number of documents. We observe an average increase of 15% in accuracy scores when we introduce relevant documents in the RAG framework. Additionally, prompt engineering guides the formulation of differential diagnosis and a chain-of-thought inference. This study is the first of its kind to identify the variation in model performance by varying the number of documents in the RAG framework
SMART: Semantic Mapping and Analysis for Regional Terrain Using Multi-Scale U-Net and TOPSIS
No full textAccurate land use classification is the backbone for urban planning. But with poor quality satellite images, varied landscapes and structures which are changing faster than ever, it becomes a challenge to define clear boundaries and hence to urban planning. This research explores the application of deep-learning model for land use classification and asses the suitability of the land. The proposed model combines a multi-scale U-Net architecture with Transformer blocks applied on a multi-spectral satellite images that improves the semantic segmentation greatly across the urban and rural regions. Additionally, a patch-wise segmentation is applied to overcome the common problem of feature imbalance and boundary ambiguity. The work extends beyond by determining the suitability scores of the regions using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) that indulges in the adaptive feature weighting. The framework has been tested on real-world datasets to evaluate the performance in classifying the land cover, that can result in actionable insights. The proposed approach when compared against existing methodologies shows significant improvements with enhanced segmentation accuracy, sharper boundary definitions, and clearer suitability mapping, achieving gains of up to 172.94% in certain land use classes. By integrating advanced deep learning techniques with geospatial decision-making models, the proposed approach offers a scalable solution for sustainable urban development