32584 research outputs found
Sort by
Real-time Adaptive Framework for Topic Modeling in Social Engineering Attacks
No full textDetecting social engineering attempts is crucial for security, as these threats are becoming more frequent and increasingly exploit human vulnerabilities. This research focuses on topic modeling using conversational data from Kevin Mitnick’s ”The Art of Deception” with dialogues that illustrate various social engineering strategies. The dataset comprises manually extracted and synthetically augmented conversations to ensure natural dialogue flow. Two methodologies are presented for utterance-level and global topic extraction: prompt engineering leveraging OpenAI’s GPT-4o-mini, characterized by few-shot learning and chain-of-thought prompting, and Quantized Low Rank Adaptation (QLoRA) utilizing Mistral’s 7B instruct model for efficient fine-tuning. Through experimentation and evaluation, this study aims to determine the effectiveness of topic modeling approaches in enhancing the identification and classification of social engineering scenario
Application of Advanced Convolutional Neural Network with Robust Hashing on Obfuscated Image Based Malware Dataset
No full textThis project report provides in-depth details on the creation of a malware classification system that makes use of Convolutional Neural Networks (CNNs) that have been strengthened by data set obfuscation and strong hashing. We test many CNN architectures, including MobileNet, ResNet, and DenseNet, using rigorous hashing and obfuscation techniques on datasets. The entire pipeline is described in this research, which ranges from the gathering and preprocessing of data sets to the application of novel hashing techniques that boost overall accuracy in classification and increase resilience against malicious attacks. Parallel to this, we show that dataset obfuscation adds an additional level of protection without significantly affecting CNNs’ capacity to pick up discriminatory characteristics. The outcomes of this research confirms that if we use robust hashing it contributes heavily on detecting the malware and improves the accuracy if desgined with advancely architected CNN models. It is practical and a secure solution for malware detection as well as classification. This work could be used as a foundation for research in the area of cybersecurity especially if we have obfuscated malware images
Transformer Integration, Fine-tuning and Zero-Shot Learning for State of Health Estimation in Li-ion Batteries using Large Language Models
No full textIn this thesis, we present a comparative analysis of the use of transformerbased and Large Language Model (LLM) models for State of Health (SoH) and Remaining Useful Life (RUL) prediction of lithium-ion batteries. With electric cars and renewable energy systems based on batteries at the forefront, the need to predict degradation accurately in order to enhance the performance and reduce maintenance costs has become imperative. Most traditional prediction methods lag the complex and non-linear characteristics of degradation in batteries, and hence the usage of sophisticated methods becomes a necessity. The research employs the CALCE dataset, which includes long-horizon cycling data for four lithium-ion batteries (CS2 35, CS2 36, CS2 37, CS2 38), to develop and evaluate two new methodologies. First, a transformer-based model is used, combining the capabilities of autoencoder features, position encoding, and multihead attentions to successfully identify temporal patterns related to battery aging. Second, a new approach is explored that takes advantage of pretraining of large language models (LLMs) via zero-shot prediction and fine-tuning methods. Visual inspection shows that both approaches skillfully capture the characteristic three-phase degradation curve of lithium-ion batteries: an initial stabilization period, a phase of gradual degradation, and a final accelerated decrease. The models successfully identify key turning points where rapid degradation begins, thus providing valuable information for maintenance policies and replacement schedules. Comparative evaluation indicates that, while transformers have more consistent behavior across different batteries, LLMs provide better uncertainty quantification and better generalization to unforeseen degradation patterns. The research highlights the need for choosing specific models to match specific application needs, with transformers performing better in some situations. This work contributes to the field by establishing a framework for advanced battery health prediction that can significantly reduce the time and resources required for degradation testing while improving the accuracy of RUL estimation in practical battery management systems
PERFORMANCE COMPARISON OF MACHINE LEARNING ACROSS METAL, CUDA, AND NEUROMORPHIC FRAMEWORKS
No full textMachine learning’s computational demands necessitate optimal performance and utilization. This research compares Apple Silicon M3 Pro with MPS, NVIDIA RTX 3070 GPU with CUDA, and neuromorphic computing for machine learning methods. We provide a cross-platform and cross-architecture performance analysis of machine learning methods to identify optimal configurations for training and inference scenarios. On traditional neural networks, Apple Silicon with MPS delivers superior energy efficiency at the cost of longer processing times for training and inference. NVIDIA with CUDA offers faster computation in training and inference at higher energy costs. Convolutional spiking neural networks perform competitively on event-based data, particularly on Apple Silicon with MPS, but underperform on frame-based data compared to traditional networks. Results highlight the importance of matching platform-architecture combinations to application requirements
Engagement Of Secondary History Social Science Teachers With Professional Development: A Phenomenological Study
Get PDFSecondary social science teachers tend to feel disengaged and frustrated with their professional development (PD) experiences. They also feel that recommended meaningful practices appear to be time-consuming, require extra hours of preparation, and rely heavily on collaboration. Using phenomenological methodology, the dissertation research investigated secondary social science teachers\u27 perceptions of their roles as educators, their professional growth, and their motivations for engaging in professional development opportunities. Nine male and one female teacher participants from a History-Social-Science Department in an urban public school in Northern California participated in an interview. This study addressed the following research questions: How do teachers perceive their growth and change as a result of participating in PD activities? What are teachers’ beliefs and perceptions about their roles as educators and their ongoing development? What factors prevent teachers from engaging in and motivating their own professional growth? The theoretical framework uses Gagné and Deci’s Self-Determination Theory of Motivation. Results indicate that teachers value peer interaction, effective PD must align with their experiences and needs, and PD should be relevant and flexible, and need to prioritize collaboration. This study supports the need to establish professional learning communities that support autonomous support, relational connection, and active competence, which could lead to more effective and transformative PD learning for all teachers and could potentially impact schoolwide changes
The Rape Kit Backlog: Problems and Solutions
Get PDFThis literature review examines the ongoing issue of rape kit backlog in the United States. Rape kits are a means of evidence collection from victims of sexual assault that can be used in court. Across the nation, many state jurisdictions have hundreds and thousands of rape kits left untested and locked away in storage facilities. Consent for testing, police bias, and funding issues are the main issues that will be addressed. Many victims choose not to have their kits tested. Other kits never get tested due to police choosing not to submit a kit due to their bias against some victims. Funding issues are the main reason why rape kits are not tested. Due to the lack of funding, other criminal cases are prioritized while rape kits are forgotten. There have been multiple suggestions to decrease the backlog: more funding to test the kits, making better rape kits, and passing legislation are some direct solutions. Spreading awareness to the community about the problem allows and encourages them to be a voice for change. When implemented, these solutions have worked, however, there are still many rape kits needing to be tested
Learning Objectives
No full textLearning Objectives is a collection of twenty-one short stories that revolve around the central theme of the loss of innocence. Through realism, fabulism, and urban fantasy, the stories examine the necessary but painful awakenings that come with maturity. These stories focus on children and the adults who choose to either protect or harm them. Set primarily within educational environments such as schools and summer camps, my stories examine the intersection of growth and disillusionment. This collection exposes the absurdities and contradictions within the education system, where miscommunication and willful ignorance can have life-long consequences. I have drawn inspiration from my own experiences working in schools and camps, where I’ve witnessed abuse and healing, trauma and resilience, and, throughout it all, absurdity and humor. My stories are not strictly autobiographical, but rather portrayals of the emotional truths that children and educators experience daily. The heightened reality of my stories explores the heightened stakes and emotions of childhood
Replicating Vignoles Et Al. Results on Multidimensional Self-construal: Comparing the United States, East Asia, Southeast Asia, and Latin America
Get PDFContemporary research on self-construal indicates that people define and perceive themselves either as independent or interdependent. Self-construal scales have become the most common method to quantitatively measure both types of self-construal across the United States, East Asia, Southeast Asia, and Latin America. Multidimensional measures of the self tend to be more accurate in defining the self than earlier dichotomous scales. In general, European Americans and Latin Americans perceived themselves as independent and East and Southeast Asians as interdependent. In the current study, undergraduate European American, East Asian American, Southeast Asian American, and Latin American students from San Jose State University were asked to complete the latest self-construal scale. This study attempted to replicate multidimensional self-construal findings on American, East Asian American, Southeast Asian American, and Latinx American samples holding acculturation constant. The results showed significant differences in harmony and self-reliance between Asian Americans and European and Latin Americans and found no significant differences between Latin Americans and European Americans. Future directions are discussed
Quantum Algorithm Emulation Using FPGAs
Get PDFField Programmable Gate Arrays (FPGAs) have been used in most of the physics sub-fields for various unique purposes. This includes particle physics, quantum optics, and, more recently, quantum computing. FPGAs boast many benefits over previous experimental and computational setups. They are versatile, easy to program, and cost-effective, leading to an understandable desire to incorporate them into the new field of quantum computing. While FPGAs have been used to help control the readout and control of physical qubits, they can also be a good tool for improving algorithm simulations, which is the focus of this paper. Different algorithms have different computational limits because of the varying operations and circuit depth levels. Due to the ease of incorporating parallel processes and the ability to optimize FPGAs on a deeper level, we can speed up computationally expensive operations, increasing the size of the quantum circuits we can simulate. This paper aims to provide an educational walkthrough for quantum computing physicists interested in taking advantage of FPGAs in their research. Our findings indicate that an inexpensive FPGA with a clock rate of 100MHz can perform simple, unoptimized matrix operations on the order of 107 ns faster than a CPU using optimized Python libraries. This can be further improved upon using more expensive hardware and optimized designs. This will provide many researchers and students with the tools needed to study quantum algorithms without the need for costly hardware
Same Face, Different Name - Colonialism, Colonization, and Capitalism: How to Combat the Insidious Capitalist Individualism that Denies Basic Human Rights to US Citizens via Community Centric Bartering
Get PDFThis paper examines the effects of colonialism, colonization, and capitalism
on the social fabric of the United States and how severing key connections
within society has hindered the direct benefits of community. As has been
tested in various cities around the world, intentional bartering can create a
sustainable community and combat the devastating effects that these
systems have had on the environment and the escalation of violence both
internally and globally. Art aimed at community rebuilding and a
decentralization of the US dollar, can help facilitate innovation that heal
various social wounds