ScholarWorks (California State University)
Not a member yet
87523 research outputs found
Sort by
Dramatic Narration in Tchaikovsky's Romeo and Juliet
No full textThis manuscript discusses the difficulties and creative possibilities of understanding Tchaikovsky's Romeo and Juliet Fantasy Overture from a conductor's perspective. By tracing the evolution of the work from its 1870 version via revisions in 1872 and 1880, this analysis shows how Tchaikovsky adapted the sonata structure and used thematic fluctuation to realize his ideas about structure and dramatic expression. Using data from previous performances, scholarly critiques, and original sources like Tchaikovsky and Mily Balakirev's correspondence, the manuscript provides a conductor's interpretation of the score as a script in which the themes inform the roles of the characters. Those conversations about elaborate rehearsal approach emphasized how they combined technical training with dramatic effect and specifically mentioned how they balanced strict performance requirements with free play. The document also discusses issues like how to work between tradition and innovation, how to maintain momentum and unity, and how to adapt the work to the size and abilities of different ensembles. It believes that the best performance of Romeo and Juliet is one that combines historical authenticity with creative stage performance, which can not only maintain Tchaikovsky's original intention, but also make modern audiences feel a superior symphony atmosphere
Framework for optimizing Network topology based on graph theory in Software-Defined Networking (SDN)
No full textThis research presents a novel innovative strategy for optimizing network topology in Software- Defined Networking (SDN) by creating a robust mathematical model for identifying crucial nodes. We introduce the Network Node Significance Index (NNSI), a new mathematical formula that quantifies node importance by integrating multiple centrality measures across different dimensions of network topology while addressing positional biases inherent in traditional methods. NNSI reorganizes centrality metrics based on their network functions (connectivity, flow control, and influence propagation) rather than their analytical scope (local, semi-local, global), providing a more intuitive and effective assessment of node criticality. This functional organization better captures how the nodes contribute to specific network operations instead of simply considering how much of the network each metric analyzes. Through extensive experimental evaluation on 261 network topologies from the Internet Topology Zoo dataset, we demonstrate that NNSI outperforms the Integrated Value of Influence (IVI) formula in identifying truly critical nodes, particularly in large, complex networks with hierarchical structures and regional clustering. Results show that NNSI identifies nodes whose removal has a 7.61% greater impact on network performance overall, with this advantage increasing dramatically to 72% in large networks exceeding 100 nodes. Building on this critical node identification capability, we develop a comprehensive topology optimization framework for SDN environments that strategically reconfigures network connections to enhance resilience, reduce latency, and improve network link utilization. Our approach leverages the centralized control capabilities of SDN to implement targeted modifications based on NNSI analysis, achieving significant improvements in network performance metrics. The proposed framework provides network administrators with a mathematically sound, computationally efficient method for identifying crucial infrastructure components and optimizing topology accordingly. This study enhances the field of network management through the progression of knowledge on node criticality within intricate networks. It showcases real-world uses for topology optimization in contemporary SDN architectures
Predictive Wear Balancing and Approximation for Efficient Non-Volatile Main Memory Management
No full textPhase-Change Memory (PCM) is a non-volatile memory technology that leverages the phase transitions of chalcogenide glass for scalable storage. While it offers low power leakage and cost-efficient read operations, it faces significant challenges due to high write energy consumption and limited endurance (107 to 109 cycles). To address these limitations, we propose a Neural Network (NN)-based data approximation framework that classifies energy levels and selectively approximates high-energy bit patterns in multi-level cell (MLC) PCM. By employing dynamic approximation techniques, our approach effectively reduces high-energy writes, leading to lower overall energy consumption. A flag byte is embedded in the least significant byte of the approximated data, enabling a Convolutional Neural Network (CNN) to accurately reconstruct the original data. This methodology enhances PCM endurance and mitigates write energy constraints while ensuring data integrity. By integrating intelligent approximation and retrieval mechanisms, our framework improves PCM's viability as a primary memory solution, addressing critical challenges in energy efficiency and longevity
SPY Stock Price Movement Technical Analysis Using Machine Learning
No full textIn the dynamic landscape of financial markets, it is crucial to have a prediction system that correctly analyzes a stock price's movement. Moreover, it is crucial to catch the shift in trends. Rather than forecasting precise price levels, this study concentrates on discerning the directional shifts in the SPY stock price. Traditional technical analysis methods, despite their widespread use, often struggle to accurately capture subtle price movements. To address these limitations, this research incorporates machine learning algorithms to develop predictive models specifically designed to identify price directionality. Using a dataset that includes historical price data and technical indicators(TIs) such as RSI, MA, MACD, and BB, the study utilizes the Random Forests algorithm to study the prediction performance. The methodology entails rigorous feature engineering, model training, and performance evaluation to build robust predictive base line models for discerning price directionality, especially reversals, along with "Up" and "Down". This research aims to fill a gap in the literature where few studies focus on reversals, providing researchers with suitable benchmark scores for future studies. For reversal prediction, the RSI + MACD + BB model emerged as the most effective, achieving the highest precision (0.52), recall (0.40), and f1-score (0.45). This result is obtained by analyzing multiple performance metrics: prediction accuracy, precision, recall, and f1 score
ANALYTICAL MODELING OF INDIUM PHOSPHIDE MESFET FOR HIGH POWER RF APPLICATIONS
No full textIndium Phosphide (InP) Metal-Semiconductor Field-Effect Transistors (MESFETs) have proven to be a potential option for millimeter-wave and high- power applications because of their excellent electron mobility and high breakdown voltage. In this work, a comprehensive analytical method is discussed for characterizing the electrical behavior of InP MESFETs, highlighting their application in RF. The model developed considers all important device parameters, like current-voltage (I-V) and capacitance-voltage (C-V) characteristics, transconductance characteristics, and noise characteristics, and it considers the gate-length modulation and series resistances effects. To be precise, the model accounts for charge partitioning in the depletion region and investigates drain current behavior in various modes of operation-linear, nonlinear, and saturation. As a main factor associated with radio frequency applications, evidence is given that transconductance and gate-source voltage are linearly related to the amplification signal gain, as well as the research also shows that the drain conductance alternately varied electric characteristics below drain-source voltages and plateaued in the high bias points region. Optimization of parasitic resistance effects is also demonstrated to significantly improve noise performance, crucial in maintaining signal clarity in high-frequency circuits. This model of analysis provides a strong foundation for the optimization of InP MESFET designs, their optimization for RF high-power applications, and the creation of future communications and radar systems. This study investigates key electrical parameters of Indium Phosphide (InP) MESFETs to evaluate their suitability for high-frequency and high-power applications. The analysis focuses on transconductance, drain-source resistance, and capacitance behavior under varying gatesource voltages and gate lengths. By examining Ids-Vds output characteristics, the work identifies optimal biasing conditions and performance trade-offs. The goal is to understand how device scaling and biasing influence efficiency, gain, and switching behavior. These insights serve to guide the design of high-speed, low-power RF circuits using InP MESFETs
Analytical simulation of ion implemented silicon carbide MESFET for post analyzing conditioner
No full textThis paper presents an analytical simulation of ion-implanted Silicon Carbide (SiC) MESFETs under post-annealing conditions. A Gaussian impurity profile is used to model the implantation process, while post-annealing is shown to significantly influence the threshold voltage, channel barrier height, and overall device performance. The study explores the impact of gate length, doping concentration, temperature, and switching dynamics on MESFET behavior. Six key MATLAB simulations are used to evaluate output characteristics, switching transients, resistance trends, temperature stability, material benchmarking, and model validation against TCAD. Results confirm that post-annealing plays a crucial role in dopant activation and channel formation, enabling optimized design for high-power and high-frequency SiC MESFET applications. These insights guide improved performance modeling and device engineering for advanced electronic systems
Electrical Design for a Smart Efficient Single-Family Residential House
No full textThe performance of electrical systems in residential buildings must prioritize safety standards and energy efficiency to achieve sustained operational effectiveness. The design and implementation of this project will establish a complete electrical system for a single-family residential house according to California Electrical Code (CEC) and National Electrical Code (NEC) requirements. The main goal is to develop an electrical layout that is safe and dependable while offering energy efficiency and meeting current power needs with possibilities for expansion. Detailed load calculations are performed to establish the main service size and determine appropriate breaker and wiring dimensions for the main service and subpanels, as well as other household equipment. The project uses AutoCAD software for precise lighting and power plans while EasyPower software develops single-line diagrams and performs voltage drop calculations and fault current analysis. The project requires solar load calculations for renewable energy integration alongside panel schedule preparation to achieve correct circuit distribution. The project maintains a comprehensive and systematic residential electrical design approach through the use of advanced design tools and compliance with established electrical codes. The suggested electrical system design focuses on ensuring user safety, promoting energy efficiency, and enabling adaptation to upcoming technological advancements. Similar residential electrical projects can utilize this study as a valuable reference that provides practical insights into the design and implementation process
Human Detection and Human Following Robot
No full textThis project focuses on the development of a human detection and following robot that combines AI-powered computer vision and real-time processing to assist in environments such as hospitals, airports, and personal care settings. Using the HuskyLens AI camera and an ESP32 microcontroller, the robot detects and tracks human faces in real time, with the ESP32 processing face position data to control motion via a PID algorithm and an L298N motor driver. Powered by a 7.4V lithium battery and equipped with wireless communication for remote updates, the robot adapts to dynamic environments and performs tasks like following individuals, delivering items, and monitoring activity. By integrating autonomous navigation with human interaction, this project presents a practical and versatile robotic solution aimed at enhancing efficiency and reducing routine human effort across various service sectors
Synthetic Data Generation from the Pediatric Heart Network Fontan I Data Set
No full textBackground: There is a growing population of patients who have undergone the Fontan surgery. Artificial intelligence models can investigate and identify trends and to predict outcomes for patients who have undergone this surgery, both in the short and the long-term. These artificial intelligence models require a substantial amount of private patient data. Generative Adversarial Networks (GANs) are a solution: they can generate an abundant amount of realistic data that satisfies the amount needed to train these artificial intelligence models. Methods: In this study, a GAN was used to generate synthetic data from the Pediatric Heart Network Fontan I Data set. Specifically, a subset containing echocardiogram and BNP measures was used to train the GAN. Two versions of this set were used: a version that has entries with missing patient information dropped, and a version that had missing information imputed using the Multiple Imputation by Chained Equations (MICE) method. Two synthetic datasets were created: one from the dataset with missing information dropped (to be referred to as synthetic-dropped), and one from the dataset with missing information imputed (to be referred to as synthetic-imputed). Both synthetic datasets were tested for fidelity using both visual (Principal Component Analysis, histograms) and quantitative (neural network / confusion matrix) methods. One neural network attempted to distinguish between real and synthetic samples, while the other was tested on its ability to predict the prevalence of ventricular dysfunction based on echocardiogram and serological information. Results: Visually (using PCA's, and histograms) both the synthetic-dropped and the synthetic-imputed datasets showed strong resemblance to their real counterparts. Quantitatively: synthetic-dropped datasets performed better in neural neural network testing. The first neural network trained on synthetic-dropped data showed a 49.3% classification error in predicting the real / synthetic status of new entries (65% error for synthetic-imputed) with a classification error of 50% representing indistinguishability between datasets. The second neural network trained on synthetic-dropped data was able to predict ventricular dysfunction in real entries with classification error of 9.44%. Conclusion: Generative Adversarial Networks can generate synthetic datasets that accurately mimic real Fontan patient data, which clinicians can use to better predict outcomes for Fontan patients
Landscape of memories: preserving oral history using volunteered geographic information (VGI) in a web GIS application
No full textThe spatial coverage of local oral history will never be a complete history; it is a patchwork at best. At the worst, it is nonexistent. One solution to achieving better spatial coverage of oral history is to offer individuals an easier means to preserve their local experiences through using volunteered geographic information (VGI) methods. VGI is geographic data that is freely shared by a person. While there are websites that allow public users to add official historical information onto maps directly, these websites do not include oral histories. This project is a proof of concept of an interactive Web GIS application that works to preserve the oral history of places far and wide, anywhere and everywhere. This is achieved by allowing public users to directly create, manage, and preserve their oral history through a spatially aware digital forum