A5087488913- Smart Agriculture and AI
- Anomaly Detection Techniques and Applications
- Remote Sensing in Agriculture
- Spectroscopy and Chemometric Analyses
- Network Security and Intrusion Detection
- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Generative Adversarial Networks and Image Synthesis
- Fault Detection and Control Systems
- Smart Grid Security and Resilience
- Industrial Vision Systems and Defect Detection
- Artificial Intelligence in Healthcare and Education
- Neural Networks and Applications
- Leaf Properties and Growth Measurement
- Machine Learning in Materials Science
- Model Reduction and Neural Networks
- nanoparticles nucleation surface interactions
- Air Quality Monitoring and Forecasting
- Remote-Sensing Image Classification
- Computational Physics and Python Applications
- Target Tracking and Data Fusion in Sensor Networks
- Advanced Image and Video Retrieval Techniques
- Advanced Combustion Engine Technologies
- Advanced Image Fusion Techniques
- Time Series Analysis and Forecasting
Massachusetts Institute of Technology
2019-2024
Iowa State University
2016-2021
Significance Plant stress identification based on visual symptoms has predominately remained a manual exercise performed by trained pathologists, primarily due to the occurrence of confounding symptoms. However, rating process is tedious, time-consuming, and suffers from inter- intrarater variabilities. Our work resolves such issues via concept explainable deep machine learning automate plant identification, classification, quantification. We construct very accurate model that can not only...
The yield of cereal crops such as sorghum ( Sorghum bicolor L. Moench) depends on the distribution crop-heads in varying branching arrangements. Therefore, counting head number per unit area is critical for plant breeders to correlate with genotypic variation a specific breeding field. However, measuring phenotypic traits manually an extremely labor-intensive process and suffers from low efficiency human errors. Moreover, almost infeasible large-scale plantations or experiments. Machine...
This paper reviews the first challenge on spectral image reconstruction from RGB images, i.e., recovery of whole-scene hyperspectral (HS) information a 3-channel image. The was divided into 2 tracks: "Clean" track sought HS noiseless images obtained known response function (representing spectrally-calibrated camera) while "Real World" challenged participants to recover cubes JPEG-compressed generated by an unknown function. To facilitate challenge, BGU Hyperspectral Image Database [4]...
Reliable seed yield estimation is an indispensable step in plant breeding programs geared towards cultivar development major row crops. The objective of this study to develop a machine learning (ML) approach adept at soybean (Glycine max L. (Merr.)) pod counting enable genotype rank prediction from in-field video data collected by ground robot. To meet goal, we developed multiview image-based framework utilizing deep architectures. Plant images captured different angles were fused estimate...
Modern distributed cyber-physical systems (CPSs) encounter a large variety of physical faults and cyber anomalies in many cases, they are vulnerable to catastrophic fault propagation scenarios due strong connectivity among the sub-systems. This paper presents new data-driven framework for system-wide anomaly detection addressing such issues. The is based on spatiotemporal feature extraction scheme built concept symbolic dynamics discovering representing causal interactions sub-systems CPS....
Modern distributed cyber-physical systems (CPSs) encounter a large variety of physical faults and cyber anomalies in many cases, they are vulnerable to catastrophic fault propagation scenarios due strong connectivity among the sub-systems. This paper presents new data-driven framework for system-wide anomaly detection addressing such issues. The is based on spatiotemporal feature extraction scheme built concept symbolic dynamics discovering representing causal interactions subsystems CPS....
This paper presents a new data-driven framework for unsupervised system-wide anomaly detection modern distributed complex systems within which there exists strong connectivity among sub-systems, operating in diverse modes and encountering large variety of anomalies. The is based on spatiotemporal feature extraction scheme built the concept symbolic dynamics discovering representing causal interactions subsystems. extracted features from pattern network (STPN) are then used to learn patterns...
Reliable training of generative adversarial networks (GANs) typically require massive datasets in order to model complicated distributions. However, several applications, samples obey invariances that are \textit{a priori} known; for example, complex physics simulations, the data universal laws encoded as well-defined mathematical equations. In this paper, we propose a new modeling approach, InvNet, can efficiently spaces with known invariances. We devise an algorithm encode them into...
Flame dynamics and combustion instability is a complex problem involving different non-linearities. Combustion has several detrimental effects on flight-propulsion structural integrity of gas turbines any such spaces where takes places internally, primarily in internal engines. The description coherent features fluid flow cases essential to our understanding the flame propagation processes. A method that able extract dynamic information from fields are generated by direct numerical...
Modern distributed cyber-physical systems (CPSs) encounter a large variety of physical faults and cyber anomalies in many cases, they are vulnerable to catastrophic fault propagation scenarios due strong connectivity among the sub-systems. This paper presents new data-driven framework for system-wide anomaly detection addressing such issues. The is based on spatiotemporal feature extraction scheme built concept symbolic dynamics discovering representing causal interactions subsystems CPS....
Deep learning models (DLMs) can achieve state of the art performance in medical image segmentation and classification tasks. However, DLMs that do not provide feedback for their predictions such as Dice coefficients (Dice) have limited deployment potential real world clinical settings. Uncertainty estimates increase trust these automated systems by identifying need further review but remain computationally prohibitive to deploy. In this study, we use a DLM with randomly initialized weights...
Deep learning (DL) models for disease classification or segmentation from medical images are increasingly trained using transfer (TL) unrelated natural world images.However, shortcomings and utility of TL specialized tasks in the imaging domain remain unknown based on assumptions that increasing training data will improve performance.We report detailed comparisons, rigorous statistical analysis comparisons widely used DL architecture binary after with ImageNet initialization (T II -models)...
Availability of an explainable deep learning model that can be applied to practical real world scenarios and in turn, consistently, rapidly accurately identify specific minute traits applicable fields biological sciences, is scarce. Here we consider one such example viz., accurate identification, classification quantification biotic abiotic stresses crop research production. Up until now, this has been predominantly done manually by visual inspection require specialized training. However,...
Advances in deep generative and density models have shown impressive capacity to model complex probability functions lower-dimensional space. Also, applying such high-dimensional image data the PDF has poor generalization, with out-of-distribution being assigned equal or higher likelihood than in-sample data. Methods deal this been proposed that deviate from a fully unsupervised approach, requiring large ensembles additional knowledge about data, not commonly available real-world. In work,...
Reliable seed yield estimation is an indispensable step in plant breeding programs geared towards cultivar development major row crops. The objective of this study to develop a machine learning (ML) approach adept at soybean [\textit{Glycine max} L. (Merr.)] pod counting enable genotype rank prediction from in-field video data collected by ground robot. To meet goal, we developed multi-view image-based framework utilizing deep architectures. Plant images captured different angles were fused...