arXiv subjects

Sarah S. Sharif

Publications and source records attributed to Sarah S. Sharif.

5 recordsLinked to original sources

arXiv2025

Accelerated Discovery of Vanadium Oxide Compositions: A WGAN-VAE Framework for Materials Design

The discovery of novel materials with tailored electronic properties is crucial for modern device technologies, but time-consuming empirical methods hamper progress. We present an inverse design framework combining an enhanced Wasserstein Generative Adversarial Network (WGAN) with a specialized Variational Autoencoder (VAE) to accelerate the discovery of stable vanadium oxide (V-O) compositions. Our approach features (1) a WGAN with integrated stability constraints and formation energy predictions, enabling direct generation of thermodynamically feasible structures, and (2) a refined VAE capturing atomic positions and lattice parameters while maintaining chemical validity. Applying this framework, we generated 451 unique V-O compositions, with 91 stable and 44 metastable under rigorous thermodynamic criteria. Notably, we uncovered several novel V2O3 configurations with formation energies below the Materials Project convex hull, revealing previously unknown stable phases. Detailed spin-polarized DFT+U calculations showed distinct electronic behaviors, including promising half-metallic characteristics. Our approach outperforms existing methods in both quality and stability, demonstrating about a 20 percent stability rate under strict criteria compared to earlier benchmarks. Additionally, phonon calculations performed on selected compositions confirm dynamic stability: minor imaginary modes at 0 K likely stem from finite-size effects or known phase transitions, suggesting that these materials remain stable or metastable in practical conditions. These findings establish our framework as a powerful tool for accelerated materials discovery and highlight promising V-O candidates for next-generation electronic devices.

cond-mat.mtrl-sci↗

arXiv2024

On-Chip Learning with Memristor-Based Neural Networks: Assessing Accuracy and Efficiency Under Device Variations, Conductance Errors, and Input Noise

This paper presents a memristor-based compute-in-memory hardware accelerator for on-chip training and inference, focusing on its accuracy and efficiency against device variations, conductance errors, and input noise. Utilizing realistic SPICE models of commercially available silver-based metal self-directed channel (M-SDC) memristors, the study incorporates inherent device non-idealities into the circuit simulations. The hardware, consisting of 30 memristors and 4 neurons, utilizes three different M-SDC structures with tungsten, chromium, and carbon media to perform binary image classification tasks. An on-chip training algorithm precisely tunes memristor conductance to achieve target weights. Results show that incorporating moderate noise (<15%) during training enhances robustness to device variations and noisy input data, achieving up to 97% accuracy despite conductance variations and input noises. The network tolerates a 10% conductance error without significant accuracy loss. Notably, omitting the initial memristor reset pulse during training considerably reduces training time and energy consumption. The hardware designed with chromium-based memristors exhibits superior performance, achieving a training time of 2.4 seconds and an energy consumption of 18.9 mJ. This research provides insights for developing robust and energy-efficient memristor-based neural networks for on-chip learning in edge applications.

cs.NE↗

arXiv2024

A Comparative Study of Sampling Methods with Cross-Validation in the FedHome Framework

This paper presents a comparative study of sampling methods within the FedHome framework, designed for personalized in-home health monitoring. FedHome leverages federated learning (FL) and generative convolutional autoencoders (GCAE) to train models on decentralized edge devices while prioritizing data privacy. A notable challenge in this domain is the class imbalance in health data, where critical events such as falls are underrepresented, adversely affecting model performance. To address this, the research evaluates six oversampling techniques using Stratified K-fold cross-validation: SMOTE, Borderline-SMOTE, Random OverSampler, SMOTE-Tomek, SVM-SMOTE, and SMOTE-ENN. These methods are tested on FedHome's public implementation over 200 training rounds with and without stratified K-fold cross-validation. The findings indicate that SMOTE-ENN achieves the most consistent test accuracy, with a standard deviation range of 0.0167-0.0176, demonstrating stable performance compared to other samplers. In contrast, SMOTE and SVM-SMOTE exhibit higher variability in performance, as reflected by their wider standard deviation ranges of 0.0157-0.0180 and 0.0155-0.0180, respectively. Similarly, the Random OverSampler method shows a significant deviation range of 0.0155-0.0176. SMOTE-Tomek, with a deviation range of 0.0160-0.0175, also shows greater stability but not as much as SMOTE-ENN. This finding highlights the potential of SMOTE-ENN to enhance the reliability and accuracy of personalized health monitoring systems within the FedHome framework.

cs.LG↗

arXiv2024

Design of Tunable Perfect Absorbers in the Mid-IR Spectrum Using Graphene-Based Multilayer Structures: Emerging Applications in Atmospheric Window Matching

This paper introduces tunable and switchable Perfect Absorbers (PAs) operating within the mid-infrared spectrum, specifically targeting the 3 to 5 um range at 0.25 um intervals. This spectrum is engineered for minimal atmospheric absorption and unique transmission characteristics. Our approach uses graphene-based nanophotonic aperiodic multilayer structures, optimized through micro-genetic algorithms within an inverse design framework. This combination broadens the design space, enabling highly accurate absorption control. Using the Transfer-Matrix-Method for simulations, we tailor absorption characteristics while keeping the structures under 2 um thick. Our results demonstrate PA tunability and switchability by adjusting graphene layers' chemical potentials. For instance, a 4 um peak can shift to 4.22 um by changing the graphene potential from 0 eV to 1 eV, without reducing efficiency. Our research also reveals PA adaptability to incident angles, maintaining 90% absorption up to 52 degrees, demonstrating versatility for applications like thermal photovoltaics, sensors, and stealth technology. This research deepens our understanding of nanophotonic materials and advances optical devices for the mid-IR range.

physics.app-ph↗

arXiv2023

Optical Properties and Behavior of Whispering Gallery Mode Resonators in Complex Microsphere Configurations: Insights for Sensing and Information Processing Applications

Whispering gallery mode (WGM) resonators are garnering significant attention due to their unique characteristics and remarkable properties. When integrated with optical sensing and processing technology, WGM resonators offer numerous advantages, including compact size, high sensitivity, rapid response, and tunability. This paper comprehensively investigates the optical properties and behavior of WGMs in complex microsphere resonator configurations. The findings underscore the potential of WGMs in sensing applications and their role in advancing future optical information processing. The study explores the impact of configuration, size, excitation, polarization, and coupling effects on the WGMs properties. The paper provides crucial insights and valuable guidance for designing and optimizing microsphere resonator systems, enabling their realization for practical applications.

physics.optics↗