Sreenidhi Institute of Science and Technology

Image-based Completely Automated Public Turing Test to Tell Computers and Humans Apart (CAPTCHA)systems are increasingly vulnerable to Adversarial Machine Learning (AML) attacks, where adversaries subtly perturb CAPTCHA images to deceive automated systems. This work proposes a novel defense mechanism that integrates Generative Adversarial Network with Convolutional Neural Network (GAN-CNN) to detect and mitigate adversarial examples. The GAN component enhances robustness by learning to generate adversarial samples during training, while the CNN is trained to classify both genuine and adversarial CAPTCHAs accurately.CAPTCHA images were obtained from the publicly available CAPTCHA Version 2 Images Dataset on Kaggle. Adversarial training, a critical step that improves the model’s resilience by introducing adversarial cases during learning, is then applied to this data. During the preprocessing phase, normalization and picture refinement techniques are used to reduce adversarial noise, blur, and distortions, guaranteeing cleaner inputs for the model.The system’s central component is a hybrid GAN-CNN architecture, in which the CNN functions as a strong feature extractor and classifier that can recognize even highly distorted images, while the GAN creates realistic adversarial CAPTCHA samples to test the classifier. First, a detection module determines if an input CAPTCHA is hostile or valid. If accepted, the image is sent to the classifier, which uses the CNN robust features to produce the final prediction. To ensure the reliability of the model and effectiveness against hostile attacks, standard metrics like accuracy, precision, recall, and F1-score are employed to evaluate the model’s general performance.

A new series of amide derivatives of oxazol-2-yl)pyrazin-2-yl)-5-(pyridin-4-yl)-1,3,4-oxadiazole derivatives were designed, synthesized and evaluated in-vitro anticancer activities against breast cancer (MCF-7), lung cancer (A549), colon cancer (Colo-205) and ovarian cancer (A2780) by using of MTT assay, and the etoposide used as reference drug. The IC50 values ranges of compound from 0.23 ± 0.045 µM to 7.38 ± 5.62 µM, where etoposide showed values ranges from 0.17 ± 0.034 µM to 3.34 ± 0.152 µM. Most of the tested derivatives were showed good to moderate activities than etoposide. This study investigates the multitarget anticancer potential of compounds 21a–21d through molecular docking and ADME–Tox analysis. The compounds demonstrated strong binding affinities and critical interactions with EGFR and VEGFR2, indicating their potential to modulate key cancer-associated pathways, including proliferation and angiogenesis. ADME–Tox predictions revealed good solubility but identified limitations such as low intestinal absorption, P-gp–mediated efflux, and inhibition of multiple CYP450 isoforms, highlighting the need for further structural optimization. Overall, these findings provide mechanistic insights supporting the potential of this scaffold in multitarget-oriented anticancer drug discovery.

A novel series of aryloxazole–1,3,5-triazin-2-yl furo[2,3-d]pyrimidine derivatives (24a–j) was synthesized and structurally characterized using appropriate analytical techniques. All compounds were evaluated for their in vitro anticancer activity against a panel of human cancer cell lines, including prostate (PC3), lung (A549), breast (MCF-7), and ovarian (A2780) cancers, using the MTT assay. The clinically established chemotherapeutic agent etoposide was employed as a positive control. Most derivatives exhibited moderate to strong cytotoxic activity compared with the reference drug. Notably, compounds 24a–24e demonstrated superior potency relative to etoposide. Among them, compound 24a, bearing a 3,4,5-trimethoxy substitution on the aryl moiety linked to the oxazole core, showed remarkable antiproliferative activity against PC3, A549, MCF-7, and A2780 cell lines, with IC₅₀ values of 1.12 ± 0.75 µM, 0.18 ± 0.046 µM, 0.23 ± 0.066 µM, and 1.09 ± 0.73 µM, respectively. Docking simulations highlighted compounds 24a and 24b as the most favorable candidates, showing strong predicted binding interactions and stable orientations within the active sites of EGFR and Aurora kinase A, with slightly stronger affinity toward Aurora kinase A. ADMET profiling indicated that most derivatives exhibit acceptable drug-like properties, including general compliance with Lipinski’s criteria and predicted good gastrointestinal absorption, with the exception of 24a. All compounds were predicted to be P-glycoprotein substrates and potential hERG II inhibitors, suggesting possible concerns related to efflux liability and cardiotoxic risk. Collectively, these results support the promise of this scaffold while emphasizing the need for further structural refinement and experimental validation to improve pharmacokinetic and safety profiles before advancing to preclinical evaluation. The compound 24a with 3,4,5-trimethoxy group bearing on the aryl moiety attached to oxazole core unit showed good activity against PC3, A549, MCF7 and A2780 cell lines with IC50 values of 1.12±0.75 μM; 0.18±0.046 μM;0.23±0.066 μM and 1.09±0.73 μM.

Frequent handovers, signaling overhead, latency, and packet loss all make it difficult to manage mobility effectively in 5G networks. These factors together compromise resource allocation, accuracy of changeover decisions, and network reliability. By integrating deep learning and meta-heuristic optimization, mobility prediction and handover performance can be improved significantly in a dynamic 5G environment. To validate this theory, this research proposes an integrated framework using Crackwave Residual Convolutional Neural Network (CRCNN) for mobility prediction and Enhanced Red Panda Optimization Algorithm (ERPOA) for optimally making handover decisions. By utilizing CRCNN, deep mobility features from users’ activity, handover requests, and real-time network variations, accurate mobility predictions are extracted and evaluated. Concurrently, by utilizing ERPOA, adaptive resource allocation helps maintain balance on cell load, reduce congestion, and minimize latency when making handover decisions through the utilization of the same computational factors used within CRCNNs. An implementation of this framework is executed in Matrix Laboratory (MATLAB) and compared against the benchmark methodologies. The results indicate that this proposed framework consistently outperformed competing methods, achieving an overall handover success rate of 98

The present era is driven by science and technology. There are many developments that are taking place on the blink of an eye every day. Of all the developments, the mode of transportation that is vehicles have marked a revolution in the history of mankind. Nowadays, transportation has become an integral part of everyone’s daily life. We are presently at a stage where every family has at least two vehicles in order to serve their daily needs. This tremendous buying and usage of the vehicles have given rise to many problems, the most serious one being traffic congestion. Currently there are many measures taken to minimize the traffic congestions but none seem too effective in controlling the traffic. In attempt to find solution to this problem, the model has been designed to deal with the problem of traffic congestion .By observing the daily scenarios, it can be inferred that the traffic congestion in a particular direction is caused recurrently during the peak hours of the day like morning and evening when people communicate to their workplaces, colleges, schools, etc., Road divider as we know serves the purpose of dividing the traffic in to and fro directions, therefore adjusting its position would serve the purpose of giving more lane space in the direction where traffic is more which would facilitate the free moment of traffic. The model is being designed using wooden plank, DC motor, limit switch, Arduino board, ESP8266, LCD, IR sensors, IC567, rack pinion, etc.