A5023319359- Adversarial Robustness in Machine Learning
- Autonomous Vehicle Technology and Safety
- Privacy-Preserving Technologies in Data
- Cryptography and Data Security
- Neural and Behavioral Psychology Studies
- Visual perception and processing mechanisms
- Security in Wireless Sensor Networks
- Traffic Prediction and Management Techniques
- Video Surveillance and Tracking Methods
- Functional Brain Connectivity Studies
- Glaucoma and retinal disorders
- Forensic and Genetic Research
- Anomaly Detection Techniques and Applications
- Network Security and Intrusion Detection
- Chaos-based Image/Signal Encryption
- Cryptographic Implementations and Security
- Advanced Malware Detection Techniques
- Transportation Planning and Optimization
- Face Recognition and Perception
- Smart Grid Energy Management
- Vehicle License Plate Recognition
- Antimicrobial agents and applications
- Visual Attention and Saliency Detection
- Bacterial biofilms and quorum sensing
- Multicomponent Synthesis of Heterocycles
Indian Institute of Technology Hyderabad
2015-2024
Karpagam Academy of Higher Education
2020
Indian Institute of Information Technology Vadodara
2018
Defence Institute of Advanced Technology
2017
Institute of Bioengineering and Nanotechnology
2017
Jawaharlal Nehru Technological University, Hyderabad
2016
Indiana University
2013
National Institute of Mental Health and Neurosciences
2013
Driving in adverse weather conditions is a key challenge for autonomous vehicles (AV). Typical scene perception models perform poorly rainy, foggy, snowy, and cloudy conditions. In addition, we observe transition states between extremes (cloudy to rainy sunny, etc.) nature with variations adversity. It crucial define understand these order develop robust AV models. Existing research works on classification focused identifying extreme However, there lack of emphasis the scenes. Hence, this...
Federated learning (FL) has emerged as a powerful machine technique that enables the development of models from decentralized data sources. However, nature FL makes it vulnerable to adversarial attacks. In this survey, we provide comprehensive overview impact malicious attacks on by covering various aspects such attack budget, visibility, and generalizability, among others. Previous surveys have primarily focused multiple types defenses but failed consider these in terms their...
Despite the high quality performance of deep neural network in real-world applications, they are susceptible to minor perturbations adversarial attacks. This is mostly undetectable human vision. The impact such attacks has become extremely detrimental autonomous vehicles with real-time "safety" concerns. black-box cause drastic misclassification critical scene elements as road signs and traffic lights leading vehicle crash into other or pedestrians. In this paper, we propose a novel...
Realization of an application using Wireless Sensor Networks (WSNs) Nodes (SNs) brings in profound advantages ad-hoc and flexible network deployments. Implementation these networks face immense challenges due to short wireless range; along with limited power, storage & computational capabilities SNs. Also, the tiny physical attributes SNs WSNs, they are prone attacks. In context attacks may range from destroying, lifting, replacing adding new The work this paper addresses threats induced...
Smart parking solution aims to output real-time occupancy information. It helps reduce bay search time, traffic, fuel consumption, and thereby vehicular emissions with increased road safety. A computer vision-based using camera video data is most reliable rational since it allows monitoring the entire open-air area at once. (cloud-based, server processing, or onboard processing) bring information end-user. comes many challenges such as viewing angles, lighting conditions, model optimization,...
Intersections are one of the main sources congestion and hence, it is important to understand traffic behavior at intersections. Particularly, in developing countries with high vehicle density, mixed type, lane-less driving behavior, difficult distinguish between congested normal behavior. In this work, we propose a way state smaller spatial regions intersections using graphs. The these graphs evolve over time reveals different states - a) forming (clumping), dispersing (unclumping), or c)...
A new sequence of pyrazoline derivatives (5a-5h) was synthesized from 2-naphthylstyryl chalcone to react with hydrazine hydrate in the presence n-butyric acid using cyclization method.Synthesized compounds chemical structure elucidated by FT-IR, Proton and Carbon NMR Spectral data CHN analysis.All were subjected in-vitro biological activity disk diffusion method.The electronwithdrawing fluoro substituted compound 5b better antibacterial against bacterial strain Pseudomonas aeruginosa also...
Parallel and automatic processing is evidenced in visual search by what commonly called popout. An object of (a target) that differs widely from all other display objects on some simple dimension a singleton; an example for red circle when displayed circles are green. A singleton attracts attention to the degree it salient, highly salient singletons produce almost independent size. The present research examines way this attraction can be diverted presence 1 or 2 nontarget perceptual...
On-demand resource provisioning is based on automated approaches for pooling and elasticity the cloud service provider (CSP) side. The infrastructure services must be adapted dynamically to accommodate customer demands yet, operate within offerings of CSP. Although multiple homogeneous clouds are available, more realistic platforms heterogeneous resources virtual machines (VMs) present unique challenges. Our management algorithm allocates memory, network computational VMs, in order provide...
Federated Learning (FL) is a collaborative learning paradigm enabling participants to collectively train shared machine model while preserving the privacy of their sensitive data. Nevertheless, inherent decentralized and data-opaque characteristics FL render its susceptibility data poisoning attacks. These attacks introduce malformed or malicious inputs during local training, subsequently influencing global resulting in erroneous predictions. Current defense strategies against either involve...