Farhan Shahriyar

Altered fingerprints are used to escape identification, so an automated system should flag them before atching. Pipelines built on handcrafted descriptors detect most altered prints but reject manygenuine ones, ndconvolutional classifiers that pool their feature maps into a single vector discard where in the print the ridge damage occurs. We propose EB2-BiLSTM, a hybrid model that reads the 9×9feature map of a fine-tuned EfficientNetB2asasequenceof81spatialpositions and encodes it with two stacked bidirectional long short-term memory (BiLSTM) layers before classification. The model is trained in two stages and selected on a validation set. On a held out test set of real prints and prints with obliteration, central rotation and z-cut alterations at three difficulty levels, EB2-BiLSTM reaches 98.58% accuracy, 98.50% specificity, a Matthews correlation coefficient of 0.9627 and an area under the receiver operating characteristic curve of 0.9983. Under an identical data split and training protocol, the stacked BiLSTM encoder raises accuracy by 0.91 points over global pooling of the same backbone and halves the number of rejected genuine prints, while attention pooling of the same positions does not give this gain. The model also exceeds a ConvNeXt-Tiny transformer hybrid that has three times as many parameters. These results show that ordered spatial context in deep features is a useful signal for screening altered fingerprints.

Quick response codes have become a common bridge between printed media and online services, and attackers now embed phishing links and malware distribution points inside them. Detection systems for this threat are usually deep networks that give a decision without any explanation, which limits trust in security operations. This paper presents an explainable hybrid quantum classical framework for malicious QR code detection. A compact convolutional neural network learns a 128 dimensional representation of each QR image. Principal component analysis compresses this representation to six components that are scaled to the interval from 0 to 𝜋 and encoded into a six qubit ZZ feature map. A fidelity quantum kernel computed from exact statevectors feeds a support vector machine with a precomputed kernel matrix. The framework is evaluated on a balanced benchmark of benign and malicious QR images across three input resolutions and three training fractions.At 128 by 128 resolution the model reaches 99.15 percent accuracy, 0.983 Matthews correlation and 0.9993 ROC AUC and it keeps 98 percent accuracy when trained on only 20 percent of the data. Five fold cross validation yields 99.22 percent mean accuracy with 0.43 percent standard deviation. Explainability is provided on both sides of the hybrid boundary through Grad-CAM saliency for the convolutional extractor and a kernel permutation importance procedure for the quantum features which shows that the first principal component drives the quantum kernel at every resolution.

The development of sustainable, cost-effective, and environmentally friendly photovoltaic absorbers is essential for advancing next-generation solar cell technologies. In this study, we investigate the thallium-based halide perovskite TlPbI₃ through a synergistic combination of density functional theory (DFT), solar cell capacitance simulator in one dimension (SCAPS-1D) device simulation, and machine learning (ML). Structural optimization, tolerance factor, formation energy, and phonon dispersion curve confirmed the structural, dynamic, thermodynamic, and mechanical stability of cubic TlPbI₃, with elastic constants fulfilling Born's stability criteria. The calculated direct band gap of 1.26 eV at the R point falls within the optimal range for single-junction photovoltaics. Charge density mapping indicated mixed ionic-covalent bonding, ensuring structural robustness. Mechanical analysis confirmed ductility and thermal stability, with an estimated melting temperature of ∼757 K. Optical results showed strong absorption in the visible region, a high static dielectric constant (ε₀ ≈ 4.6), and low reflectivity, underlining the suitability of TlPbI₃ for optoelectronic devices. SCAPS-1D simulations were performed on different heterojunction configurations, optimizing absorber thickness, doping density, defect density, and buffer layers. The best-performing device, fluorine-doped tin oxide (FTO)/cadmium sulfide (CdS)/thallium lead triiodide (TlPbI₃)/copper (Cu), delivered a power conversion efficiency (PCE) of 22.20% with open-circuit voltage (V_OC) = 0.7987 V, short-circuit current density (J_SC) = 34.56 mA cm⁻², and fill factor (FF) = 80.40%, confirming the strong photovoltaic potential of TlPbI₃. Machine learning models, trained on simulation datasets, successfully identified absorber thickness and defect density as the most critical factors influencing device performance. This integrated computational framework demonstrates the potential of TlPbI₃ as a viable absorber material for next-generation solar cells and provides valuable predictive insights to guide experimental development.

Quantum computing threatens classical cryptography and the convergence of quantum key distribution (QKD), and steganography offers an unconditionally secure hybrid approach for postquantum communication. QKD ensures provable secrecy via quantum mechanics, while steganography conceals the existence of communication providing stealth and plausible deniability. This systematic review explores integrated frameworks that combine QKD with classical and quantum steganographic techniques to enhance confidentiality, integrity, and undetectability. The study categorizes existing hybrid models and evaluates their performance using metrics such as quantum bit error rate (QBER), entropy, and imperceptibility. It examines quantum steganographic protocols including quantum LSB, entangled‐state hiding, and covert key transmission, analyzing their resilience against quantum noise and steganalysis. Despite promising simulation tools and real‐time QKD implementations, challenges remain in scalability, interoperability, and formal security guarantees. This review identifies key limitations in current systems and highlights future directions for developing robust, scalable, and stealth‐aware quantum–classical security frameworks. To the best of current knowledge, this is the first systematic synthesis of hybrid QKD–steganography frameworks offering a novel taxonomy and critical analysis to advance secure communication infrastructures in the postquantum era.

Smishing is a social engineering attack using SMS containing malicious content to deceive individuals into disclosing sensitive information or transferring money to cybercriminals. Smishing attacks have surged by 328%, posing a major threat to mobile users, with losses exceeding \$54.2 million in 2019. Despite its growing prevalence, the issue remains significantly under-addressed. This paper presents a novel hybrid machine learning model for detecting Bangla smishing texts, combining Bidirectional Encoder Representations from Transformers (BERT) with Convolutional Neural Networks (CNNs) for enhanced character-level analysis. Our model addresses multi-class classification by distinguishing between Normal, Promotional, and Smishing SMS. Unlike traditional binary classification methods, our approach integrates BERT's contextual embeddings with CNN's character-level features, improving detection accuracy. Enhanced by an attention mechanism, the model effectively prioritizes crucial text segments. Our model achieves 98.47% accuracy, outperforming traditional classifiers, with high precision and recall in Smishing detection, and strong performance across all categories.