SHSushmit Hossain

Sushmit Hossain

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sushmit Hossain has a total of 37 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Inverse Design of Optical Color Routers with Improved Fabrication Compatibility

    by Sushmit Hossain, Zerui Liu, Nishat Tasnim Hiramony, Tinghao Hsu, Himaddri Roy, Hongming Zhang, Wei Wu

    We present a Genetic Algorithm (GA)-based inverse design framework for creating a single-layer, fabrication-compatible dielectric nano-patterned surface that enables efficient color routing in both transmissive and reflective optical systems. Unlike traditional multilayer or absorption-based color filters, the proposed structure employs a fabrication-compatible architecture that spatially routes red, green, and blue light into designated output channels, significantly enhancing light utilization and color fidelity. The design process integrates a GA with full-wave finite-difference time-domain (FDTD) simulations to optimize the structural pillar height distribution, using a figure of merit that simultaneously maximizes optical efficiency and minimizes spectral crosstalk. For CMOS image sensor-scale designs, the nano-patterned surface achieved peak optical efficiencies of 76%, 72%, and 78% for blue, green, and red channels, respectively, with an average efficiency of 75.5%. Parametric studies further revealed the dependence of performance on pillar geometry, refractive index, and unit cell scaling, providing practical design insights for scalable fabrication using nanoimprint or grayscale lithography. Extending the approach to reflective displays, we demonstrate tunable-mirror-based architectures that emulate electrophoretic microcapsules, achieving efficient color reflection and an expanded color gamut beyond the sRGB standard. This single-layer, inverse-designed nano-patterned surface offers a high-performance and fabrication-ready solution for compact, energy-efficient imaging and display technologies.


  • A Stochastic Analog Boolean Satisfiability Solver

    by Shiyu Su, Qiaochu Zhang, Zerui Liu, Hsiang-Chun Cheng, Zhengyi Qiu, Mayank Palaria, Jiacheng Ye, Deming Meng, Buyun Chen, Sushmit Hossain, et al.

    This article presents a stochastic analog Boolean satisfiability (SAT) solver, featuring a fast open-loop architecture with continuous-time (CT) self-loopback pull-up switches, a discrete-time (DT) scrambling scheme, and a cost-efficient hybrid random code generator. The SAT prototype demonstrates 100% solvability and 3.5- $\mu $ s solution time with 8.6-nJ energy consumption for 1000 hard benchmark problems (20 variables and 91 clauses) in 65-nm complementary metal–oxide–semiconductor (CMOS), achieving over $1000\times $ improvement in solution time compared to the prior analog SAT solver, and more than $10\times $ improvement compared to state-of-the-art digital SAT solvers implemented in a similar process without additional pre-processing. Moreover, the proposed SAT solver is highly flexible and modular, allowing a low-complexity, low-cost, and scalable design.


  • Functionalized Nanofinger Enhances Pretrained Language Model Performance for Ultrafast Early Warning of Heart Attacks

    by Hongming Zhang, Zerui Liu, Heming Sun, Yunxiang Wang, Ting-Hao Hsu, Sushmit Hossain, Nishat Tasnim Hiramony, Himaddri Roy, Hao Zhou, Matthew Tan, et al.

    Heart attacks are the leading cause of death worldwide, which means an accurate early warning system is needed. Traditional methods, such as an electrocardiogram (ECG) and blood test, usually require expert interpretation and take more than 15 min to obtain diagnostic results, which often results in delayed treatment. Our previous work [Liu, Z. Small 2023, 19(2), e2204719] developed a functionalized nanofinger platform utilizing Raman spectroscopy and machine learning (ML) to detect heart attack-related biomarkers, brain natriuretic peptide (BNP) in blood samples, achieving a high true positive rate (98%) but with limited true negative accuracy. This study enhances diagnostic accuracy by integrating the AI-driven analysis of patient symptoms with blood biomarker analysis. We incorporate additional patient features, including current and historical symptoms. We fused the biomarker test results, which is a one-dimensional probability vector ranging from 0 to 1, with patient symptom description as part of the sentence, which is then processed by a fine-tuned pretrained language model to generate embeddings and passed to a classification head for diagnosis. For data augmentation, we employ generative models to synthesize realistic patient cases, expanding the data set and improving model robustness. Our approach achieves an accuracy rate of 99.19%, outperforming conventional diagnostic methods. Given the difficulty of obtaining sufficient clinical data, this study presents a scalable AI-based solution for early-stage heart attack detection by integrating functionalized nanofinger-based blood analysis with large language models (LLMs) and leveraging generative models to synthesize realistic patient cases for data augmentation. This approach significantly advances the automation of diagnosis in modern healthcare and can be readily adapted to other disease detection tasks.


  • Open Access

    End-to-End Diverse Metasurface Design and Evaluation Using an Invertible Neural Network

    by Yunxiang Wang, Ziyuan Yang, Pan Hu, Sushmit Hossain, Zerui Liu, Tse-Hsien Ou, Jiacheng Ye, Wei Wu

    Employing deep learning models to design high-performance metasurfaces has garnered significant attention due to its potential benefits in terms of accuracy and efficiency. A deep learning-based metasurface design framework typically comprises a forward prediction path for predicting optical responses and a backward retrieval path for generating geometrical configurations. In the forward design path, a specific geometrical configuration corresponds to a unique optical response. However, in the inverse design path, a single performance metric can correspond to multiple potential designs. This one-to-many mapping poses a significant challenge for deep learning models and can potentially impede their performance. Although representing the inverse path as a probabilistic distribution is a widely adopted method for tackling this problem, accurately capturing the posterior distribution to encompass all potential solutions remains an ongoing challenge. Furthermore, in most pioneering works, the forward and backward paths are captured using separate models. However, the knowledge acquired from the forward path does not contribute to the training of the backward model. This separation of models adds complexity to the system and can hinder the overall efficiency and effectiveness of the design framework. Here, we utilized an invertible neural network (INN) to simultaneously model both the forward and inverse process. Unlike other frameworks, INN focuses on the forward process and implicitly captures a probabilistic model for the inverse process. Given a specific optical response, the INN enables the recovery of the complete posterior over the parameter space. This capability allows for the generation of novel designs that are not present in the training data. Through the integration of the INN with the angular spectrum method, we have developed an efficient and automated end-to-end metasurface design and evaluation framework. This novel approach eliminates the need for human intervention and significantly speeds up the design process. Utilizing this advanced framework, we have effectively designed high-efficiency metalenses and dual-polarization metasurface holograms. This approach extends beyond dielectric metasurface design, serving as a general method for modeling optical inverse design problems in diverse optical fields.


  • Open Access

    Plasmon-Enhanced Photocatalytic CO2 Reduction for Higher-Order Hydrocarbon Generation Using Plasmonic Nano-Finger Arrays

    by Tse-Hsien Ou, Pan Hu, Zerui Liu, Yunxiang Wang, Sushmit Hossain, Deming Meng, Yudi Shi, Sonia Zhang, Boxin Zhang, Boxiang Song, et al.

    The carbon dioxide reduction reaction (CO2RR) is a promising method to both reduce greenhouse gas carbon dioxide (CO2) concentrations and provide an alternative to fossil fuel by converting water and CO2 into high-energy-density chemicals. Nevertheless, the CO2RR suffers from high chemical reaction barriers and low selectivity. Here we demonstrate that 4 nm gap plasmonic nano-finger arrays provide a reliable and repeatable plasmon-resonant photocatalyst for multiple-electrons reactions: the CO2RR to generate higher-order hydrocarbons. Electromagnetics simulation shows that hot spots with 10,000 light intensity enhancement can be achieved using nano-gap fingers under a resonant wavelength of 638 nm. From cryogenic 1H-NMR spectra, formic acid and acetic acid productions are observed with a nano-fingers array sample. After 1 h laser irradiation, we only observe the generation of formic acid in the liquid solution. While increasing the laser irradiation period, we observe both formic and acetic acid in the liquid solution. We also observe that laser irradiation at different wavelengths significantly affected the generation of formic acid and acetic acid. The ratio, 2.29, of the product concentration generated at the resonant wavelength 638 nm and the non-resonant wavelength 405 nm is close to the ratio, 4.93, of the generated hot electrons inside the TiO2 layer at different wavelengths from the electromagnetics simulation. This shows that product generation is related to the strength of localized electric fields.