Evaluating Generalizability of Deep Learning Models Using Indian-COVID-19 CT Dataset
Suba S,Nita Parekh,Ramesh Loganathan,Vikram Pudi,Chinnababu Sunkavalli
International Conference on Bioinformatics and Data Science, ICBDS, 2023
@inproceedings{bib_Eval_2023, AUTHOR = {Suba S, Nita Parekh, Ramesh Loganathan, Vikram Pudi, Chinnababu Sunkavalli}, TITLE = {Evaluating Generalizability of Deep Learning Models Using Indian-COVID-19 CT Dataset}, BOOKTITLE = {International Conference on Bioinformatics and Data Science}. YEAR = {2023}}
Computer tomography (CT) have been routinely used for the diagnosis of lung diseases and recently, during the pandemic, for detecting the infectivity and severity of COVID-19 disease. One of the major concerns in using machine learning (ML) approaches for automatic processing of CT scan images in clinical setting is that these methods are trained on limited and biased subsets of publicly available COVID-19 data. This has raised concerns regarding the generalizability of these models on external datasets, not seen by the model during training. To address some of these issues, in this work CT scan images from confirmed COVID-19 data obtained from one of the largest public repositories, COVIDx CT 2A were used for training and internal validation of machine learning models. For the external validation we generated Indian-COVID-19 CT dataset, an open-source repository containing 3D CT volumes and 12096 chest CT images from 288 COVID-19 patients from India. Comparative performance evaluation of four state-of-the-art machine learning models, viz., a lightweight convolutional neural network (CNN), and three other CNN based deep learning (DL) models such as VGG-16, ResNet-50 and Inception-v3 in classifying CT images into three classes, viz., normal, non-covid pneumonia, and COVID-19 is carried out on these two datasets. Our analysis showed that the performance of all the models is comparable on the hold-out COVIDx CT 2A test set with 90%–99% accuracies (96% for CNN), while on the external Indian-COVID-19 CT dataset a drop in the performance is observed for all the models (8%–19%). The traditional machine learning model, CNN performed the best on the external dataset (accuracy 88%) in comparison to the deep learning models, indicating that a lightweight CNN is better generalizable on unseen data. The data and code are made available at https://github.com/aleesuss/c19.
Evaluating Generalizability of Deep Learning Models Using Indian-COVID-19 CT Dataset
Suba S,Nita Parekh,Ramesh Loganathan,Vikram Pudi,Chinnababu Sunkavalli
International Conference on Bioinformatics and Data Science, ICBDS, 2023
@inproceedings{bib_Eval_2023, AUTHOR = {Suba S, Nita Parekh, Ramesh Loganathan, Vikram Pudi, Chinnababu Sunkavalli}, TITLE = {Evaluating Generalizability of Deep Learning Models Using Indian-COVID-19 CT Dataset}, BOOKTITLE = {International Conference on Bioinformatics and Data Science}. YEAR = {2023}}
Computer tomography (CT) have been routinely used for the diagnosis of lung diseases and recently, during the pandemic, for detecting the infectivity and severity of COVID-19 disease. One of the major concerns in using ma-chine learning (ML) approaches for automatic processing of CT scan images in clinical setting is that these methods are trained on limited and biased sub-sets of publicly available COVID-19 data. This has raised concerns regarding the generalizability of these models on external datasets, not seen by the model during training. To address some of these issues, in this work CT scan images from confirmed COVID-19 data obtained from one of the largest public repositories, COVIDx CT 2A were used for training and internal vali-dation of machine learning models. For the external validation we generated Indian-COVID-19 CT dataset, an open-source repository containing 3D CT volumes and 12096 chest CT images from 288 COVID-19 patients from In-dia. Comparative performance evaluation of four state-of-the-art machine learning models, viz., a lightweight convolutional neural network (CNN), and three other CNN based deep learning (DL) models such as VGG-16, ResNet-50 and Inception-v3 in classifying CT images into three classes, viz., normal, non-covid pneumonia, and COVID-19 is carried out on these two datasets. Our analysis showed that the performance of all the models is comparable on the hold-out COVIDx CT 2A test set with 90% - 99% accuracies (96% for CNN), while on the external Indian-COVID-19 CT dataset a drop in the performance is observed for all the models (8% - 19%). The traditional ma-chine
Understanding researcher’s role in enabling or inhibiting innovation clusters in emerging Indian economy.
SWARAJ SINGH CHAUHAN,Ramesh Loganathan,Nimmi Rangaswamy
R&D Management Conference, R&D MC, 2022
@inproceedings{bib_Unde_2022, AUTHOR = {SWARAJ SINGH CHAUHAN, Ramesh Loganathan, Nimmi Rangaswamy}, TITLE = {Understanding researcher’s role in enabling or inhibiting innovation clusters in emerging Indian economy.}, BOOKTITLE = {R&D Management Conference}. YEAR = {2022}}
This paper presents a descriptive analysis to
understand how researchers play a significant role within
emerging innovation clusters. We highlight reasons that
enable or inhibit researchers towards productization within scientific institutes in India. The paper focuses on
understanding the above from a researcher’s point of view by analysing their motivation and concerns.
Most research work from deep-tech computer science
domains contain a shelf life when it comes to productization. Therefore, associated activities have a designated time period related to productization to avoid research leading to the product from falling into the Valley of death.
As a result, researchers who work on projects resulting in potential productization become a crucial part of a successful research to productization flow of work.
Our paper develops an understanding of and evaluates
different aspects of a researchers’ journey as they relate to productization of deep-tech Computer Science research work.
Index Terms - Innovation Clusters, Productization, Emerging Economy, Deep-tech Computer Science, India
Framework for Automated Attendance & Attention Tracking to Address Learning Gaps Due to Pandemic
Pranavi Pendyala,Sriya Reddi,Arjun Rajasekhar,Syed Falahuddin Quadri,N. Jaisankar,Ramesh Loganathan
International Conference on Teaching, Assessment and Learning for Engineering, TALE, 2022
Abs | | bib Tex
@inproceedings{bib_Fram_2022, AUTHOR = {Pranavi Pendyala, Sriya Reddi, Arjun Rajasekhar, Syed Falahuddin Quadri, N. Jaisankar, Ramesh Loganathan}, TITLE = {Framework for Automated Attendance & Attention Tracking to Address Learning Gaps Due to Pandemic}, BOOKTITLE = {International Conference on Teaching, Assessment and Learning for Engineering}. YEAR = {2022}}
The sudden shift to online education due to the pandemic has led to widespread learning gaps in the student population, increasing the need for individualistic attention to student learning as we transition back to physical classrooms. One can take advantage of the advancements in technology to help in this process by developing tools that can assist teachers decrease the amount of administrative workload and improve their ability to track the learning and understanding of every student in the classroom. The current work proposes a framework for accurate, low-cost, and scalable computer visionbased attendance and attention tracking in physical classroom environment to tackle this problem