Good morning Sir. I am Doctor G Nirmala working as an associate professor currently in RMD Engineering College. And I have applied for this because I am interested in the institute I was the following year in the LinkedIn. For VIT and having the aspiration to join in your institute.

Thank you for sharing that, Professor.

I'm Doctor Jean Nirmala, working as associate professor currently. I joined this profession as a lecturer in 2003. And I am. I'm in 23 years of experience. And. I was worked in various departments. I was appointed as a. As an as a lecturer in MC department and since I have completed MCA ME PhD under university. I have completed my PhD in Information Communication Engineering in 2002. In Anna University, I have completed my computer science and engineering under Anna University in the RMD engineering colleges part-time engineering faculty I worked. And.

I'm pleased to attend this interview.

Thank you for outlining your career path, Professor.

Thank you for that overview, Professor. Reflecting on your teaching career, is there a particular foundational topic in computer science or information communication engineering that you find especially rewarding to introduce to first-year students?

Object oriented concepts. In using C++, it is a foundation course. It is introduced in all the first year curriculum.

To impart the foundations. How to use the object? How to use the functions repeatedly?

Thank you, Professor. Your recent publication on “An ensemble learning approach for intrusion detection in IoT-based smart cities” is quite relevant to our department’s interests. Could you walk me through the core methodology of that work, and how you might translate those technical concepts into material that undergraduate students could grasp?

You mentioned teaching object-oriented concepts using C++ to first-year students—can you share a specific classroom strategy or example you use to help students understand and apply the idea of encapsulation effectively?

Encapsulation is defined as a wrapping up of data. We want to hide the data and we it is called a generalization. We can implement the. Function definition and we can hide the data because data is very important, we have to provide the security for that. What is the classroom strategy which I have followed? 1st I used to explain the. Uh, need of the encapsulation and I used some uh, like pedagogy method I can apply so that they can clear and understand that I can form uh, in the students, uh, as a real time example with a queue. And how to protect a particular girl and we can call the method. So this is actually encapsulation is hiding. It is used for hiding to protect the data. So we can explain with the real time example along with the code. Before explaining the code we can use this strategy so that they can understand very easily. They should be very strong in foundation then only they can. Follow the concept and they can reply with the code. They can then write the code to pass the test cases. Uh. Then, uh, they we will get a success.

You mentioned using real-life analogies and code demonstrations before students write their own code—can you recall a specific instance where this approach helped a struggling student grasp encapsulation, and what was the outcome?

We can give a real time analogy that we we can form a set of students as a different function and in that set of students we can we can call a particular student and we can wrap the. Particular student and then we can call the other students to. Have a function call. Uh, then we can make. The particular students to disclose so so the so that they can struggle what the strap. Bing which data is wrapped so they can understand when we form a student's queue and we can represent real time? Then we can make them to understand who struggle. For that they need generally a real time example along with the. The students forming as a role play model. We can use a role play model for explaining this concept in so that they can have a thorough understanding. Understanding rather than writing the code directly and calling the functions.

Thank you for describing your role-play approach. Shifting to your research, can you discuss how your work in information communication engineering has influenced the way you design laboratory sessions or student projects, particularly in multimedia or AI in media applications?

In Information Communication engineering, actually I have completed me computer science so I was registered. Under the Information Communication Engineering, my research basically concentrates on machine learning. I started my research at 2015. I completed UH-4. Papers subject subject related to advanced digital signal processing then mathematics paper then followed by the basics required for the research. After that I have completed the confirmation meeting in that what is my research topic is the image forgery deduction using CNN. That is my research in that I have applied for three different papers. 1st paper was finding out the similarity. In the image 4G. When I take the forged images and find out the correlation between the images, then I use the statistical measures like mean and median and applied for the conference. So it was published in Scopus and it was cited by many of the core scholars, other research scholars. So after completing that particular paper in the conference, then I have presented in the confirmation meeting. Then I continued by research in applying the machine learning after. In that. The image for images are collected from the kaggle database then the forgery among the images whether there are some changes because it is the need of the hour we find out the forgery. This is one of the important topic coming under image 4:00 and 6:00. So under that I have collected images in the kaggle which is a publicly available database then I have. Worked on that. I find out using the machine learning algorithm like I used particularly Support Vector Machines to classify the margins and I clearly updated along with the machine learning algorithm I used.

CNN, CNN method because what is the difference that is coming under deep learning? The difference between the machine learning and the deep learning is where I found the changes. I need some pre processing which is taking a lot of time. Equally we have to make it and we have to upload the image to the machine learning algorithm whereas in the deep learning it is not required. In deep learning CNN automatically if we include the images then it can extract and it can find out directly. The CNN classifiers I mean with the layers. Then it can easily say where is the forgery and whether it happened or not. 1st I used binary classifiers to find out whether it is there or not. Then second where the segmentation I applied, where is the exact forgery applied uh or changed. Then we used uh uh to enhance the performance. Again used uh classification method. Then I got the result. So 1st paper is related to machine learning. Then second paper is uh using the convolution neural network. Then I returned the thesis and presented UH. Then I got awarded UH in the PST. The major area is machine learning and the deep learning.

In the paper I want to use.

You described using Support Vector Machines and CNNs for image forgery detection—can you give a concrete example of how you translated your research findings from these projects into hands-on lab exercises or student project topics?

You described using Support Vector Machines and CNNs for image forgery detection—can you give a concrete example of how you translated your research findings from these projects into hands-on lab exercises or student project topics?

Thank you for sharing those details. Now, looking ahead, could you outline your three-year research roadmap and explain how you plan to align your research focus with the goals of an institution like VIT, particularly with regard to interdisciplinary collaborations or securing external funding?

So in my research journey after completing the PhD, I applied interdisciplinary field like we can take the data by using the medical imaging. This CNN classification is particularly used in a disease classification and identification. We can use uh, I, I have uh, used my aim uh, regarding, uh, with your VIT uh. Inspirational goals are it is providing a research environment. Now currently I'm working under autonomous institution and I'm the branch coordinator for the past 10 years. So I was supposed to concentrate on academics work and I'm unable to concentrate on the research work. So I want to align myself towards the VIT research careers expecting that I have applied for that and. Interdisciplinary field. I want to apply machine learning, deep learning techniques which I have already having experience. In the medical field and I want to detect some wearable devices like which is used for the patients faster recovery and I want to contribute myself in the research perspective. Towards the medical field which is aiming or helping the patients as well as doctor for faster recovery as well as faster disease identification based on the road map which I'm going to. Planning to work in these interdisciplinary field.

Thank you, Professor. Now, for a large-enrollment course where slides are not permitted, how would you implement an active learning approach—such as a flipped classroom or peer instruction model—to keep students engaged and ensure deep understanding of the material?

Now what is actually the need of the hour is we require some. Active listening approaches because students lagging concentration, they are not enabled to they are not able to concentrate for continuously for 5 minutes or three minutes. When we say literally the text. Or when we follow the older method. Then it is not that much effective because the uh. The concepts that clearly we used some active learning approaches and using currently using the applications like. We have developed a based application which we will once we talk the topic. We used. I used to have a chalk and board method to. Uh, introduce the topic. And what are the categories that is coming under that then I should I will use in the? Projects and where it is really implemented and then, uh, the concepts we can use with the help of a PowerPoint presentation in a. Uh, sliding board we can use, uh, I'm using then followed by. After the completion of that, I used to have some objective tests, uh, using the. Uh, next Gen. app we are using the application and in the active, uh classrooms we used uh direct sessions. Umm, like, uh. After completing the class, we will make the students currently to display in the board. Using that application, we will have a active assessments online assessments immediately after completing the topic to have 5 to 10 questions to assess the understanding level using the MCQS objectives. Then we can make the concepts very clearly, particularly by using a Google form. Or we can have active learning app we are using for conducting the assessments. And subjective assessments for practicing we use it is actually followed currently with the flipped classrooms and the A based classrooms we can impart the concepts very clearly rather than 23 years back. We use only chalk to represent diagrams. Requires more time to draw and it is not effective method which we are implementing currently. Using the a now we are adapting to that to vastly complete the. The bus as well as make them to contribute and understand the concepts implement in some pattern. Enter uh, making it as a research paper. Uh, we are using uh currently.

Let’s discuss your engagement with departmental administration. Suppose your department at VIT needs to improve its research output per faculty member to meet upcoming accreditation requirements. What systematic steps would you take to help raise this metric?

Uh, in the.

Suppose your department at VIT needs to improve its research output per faculty member to meet upcoming accreditation requirements. What systematic process would you implement to track, motivate, and document faculty research contributions throughout the year?

In the departmental agree. I am the NBA in charge for the 5th criteria. It is a faculty contribution. We are alerted with different criteria coordinators. Under that in the 5th criteria, my work is on faculty patent publication, research publications, their contribution, their date of joining, so the entire database regarding the department management along with the applications where we are, whatever we are using. To faster the counseling methods to impart the mentoring skills I used to contribute in the NBA and I have attended 4 different Nbas in my tenure in the R&D engineering college. For the past 20 years, I have presented in front of the NBA team to represent our college. Criteria For the criteria, what are the faculty appraisals we have given? What are the salary details we have given? What are their? Objectives We have represented their contribution in terms of conferences, papers, their contribution towards the outside world and the teaching learning process is coming under the second criteria. Now I am responsible for the criteria. Currently we are also going for the. 5th time NBA this time and I am taking care of the faculty guest lectures and the contributions faculty are doing to the other institution. To impart their knowledge faculty FDPC organized these files and all are coming under me and I am categorizing and I'm completely coordinating the 5th criteria I'm I will represent on behalf of the department to the NBA team. To represent uh. Uh, the. Features of our department.

Given your experience as NBA coordinator for faculty contributions, what specific changes or digital tools have you introduced to streamline the documentation process for accreditation audits, and what impact did those changes have?

Digital tools I have used. To receive the data. Data collection plays a vital role in the NBA, right in the NBA submissions. So the evaluation is based on the data that is available if we, if we prove, if we provide the data. Belonging to that particular department the collection data collection plays a vital role you for that we have. I have used the online tools like cloud. Using the cloud I used. Google forms that is we are using as a software services using that we receive the data like they are professional society contribution, they are professional membership then what are their conferences attended along with the proof with the certificates that you received. Then we used Google Drive using that we can store make them to store it under particular name. Then we can form a table for SAR for. Particular table in the SAR we can have a generate the proof for that. The software mainly used is. I have conducted using the application called as Pragadi app. Inside that the faculty details are also available along with the counseling each student. Is allocated to a particular counselor for one staff member will be having a 22 or 23 councillors and the students they can have a mentoring particularly after the completion of internal assessments. We will store this data these central when we are recording in hand or when we are recording by using simple task, it is very difficult to maintain. So what we used is we used application like. Pragati app, a separate application for online counseling in RMK. Particularly we are using to record the interaction with the parents. What is the progress related to that? So this is coming under the usage of application which is. Mainly. Plays a helpful. In a data collection and retrieval we will make easily the faster the entries in the SAR submission and we use. Generally the applications are used is using the Pragati app then using the Google. Form Google Sheets we use to share. The data collection I'm using by cloud applications as the software services we used to not able to interact with the staff so we can get offline and they can upload everything from their home and we can use it for evaluation.

Thank you for that explanation. Now, imagine a situation where a student alleges that your grading was biased, but the department is also under pressure to meet a specific pass-percentage target for accreditation. How would you approach resolving this conflict while maintaining both academic integrity and institutional requirements?

Actually in the past 23 years in my journey. I will teach every student as a equal person only I I am not using any bias while grading the assessments, whatever it is available in the paper, whatever they have completed in the testing. We used to put. Then generally like the institution, like us, like private colleges, all these we are in pressure. So what we will do is we will have some boosting methodology based on the track record they are having. To improve the. Marks then we will use another important method is we will have a remedial classes. On the evening coaching classes, particularly for the hostel students to make those students. To concentrate more and separate individual concentration is made for the weak students. So for weak students, we will identify in the beginning itself. Once they're tested, one of the assessments is completed, we will sort out who are the top students, who are the weak students, who are the media per students and based on the weak students strategy, what is the difficulty they are facing. We will interact with them. Particularly we will allocate the evening classes and remedial classes to improve the percentage because the pressure is we have to improve the percentage and the outcome is very important. We will. We have to impress the students we have. To implement the concepts. Teach with them separately. If we individually concentrate which each and every student, the follow plays a vital role if we follow but definitely there will be outcome. The counseling and the follow-up is one of the important criteria for the students to up. Lifted themselves in the weaker areas. This also lies in the hands of the faculty. We will definitely achieve the required output if we have a remedial. Classes and particularly the hostel students week students, we will concentrate with them and reach the outward. That confidence we will have and I I have applied this methodology for the past 23 years, so I am damn sure that we can get the required outcome by specific concentration. Towards the big students.

Let’s turn to industry alignment. With your background in machine learning and digital signal processing, what concrete steps would you take to leverage your professional network or industry contacts to establish internship or employment pipelines for students at VIT?

For professional network generally I use LinkedIn. Uh, in the LinkedIn. Website we will have a idea of where and all what and all going on the programs and how can you interact with the industry Center for doing some certifications. We will follow up the required groups. Like a machine learning groups, Deep learning groups, CNN group. If we follow those type of groups then we will have a strong. Communication at. Networking with the friends who are all doing the. Researchers in the multidisciplinary fields and we will form a separate group. We can apply, we can implement work towards that along with the two students team and we can reach the target. Uh, particularly the currently working professional network, uh, we are using is a lingering. Which makes a very good. Uh, awareness for what and all going on in all our institution all over, uh, the world. What are the certifications required and what are the offers that is available You can. Read that and you can have a strong network with the LinkedIn.

Understood. Let’s discuss your expertise in multimedia or AI in media—can you describe a specific project or application where you developed or applied these techniques, and explain the technical challenges you addressed?

Shortest route in the road when we are. Following uh, the rules via, uh, Google map or something, we will show what is actually, uh, red color means more traffic. And the green color means no traffic and orange color means mild traffic. So to understand during some emergency purpose particularly that application is developed. To root immediately. Faster so in that what is the difficulty I have faced? Is the a which capture video capturing is there is to be happened during the? Travel way and generally this applications are particularly useful in. Rapido or some called taxis. So when we are in the transport we require when particular political meeting is happening on then the routes will be diverted. So the aim of the project is to fastly represent the. Alternate route or emergency purpose. We developed the application and we have. Represented. Uh, in the. Presentation. Uh, then I have, uh, told the students to. Improve uh. And Fastly developers are complete application along with my mentor. We are using that and currently I'm deploying that application. In that the process is going and the presentation and the review 3 reviews are completed. We have given suggestions then and there and we are implementing. The project when compared to the older method, the main contribution is a faster representation for the alternate route and emergency. Transports should be given more priority. Uh, that is the aim of uh, the project. Lastly, we have implemented using the. My students, along with the team of our students. Have represented.

Thanks for explaining the project. When your team integrated live video capture for traffic analysis, what specific data processing or machine learning techniques did you use to extract real-time traffic conditions from the footage?

CNN, particularly the images or video capturing using the. Uh, we have a find out the the. It is actually not the machine learning method. The advanced way of machine learning is a deep learning because the. Uh, live video capture, uh, we have received it through the. Surveillance camera. The traffic signals what we are using. So we have taken the video and extracted as a various trips and find out. Uh, split the video into various images and uh, we have find out, uh, the similarity in that the Vata, the artery troops, uh, particularly image, uh, classification we used in the CNN for finding out the traffic. Under the representation in the form of a color representations in the dashboards. Particularly it is used in the display in the dashboards. Uh, I have used the technique.

Understood. When you used CNNs to classify the extracted images for traffic density, how did you label or validate the training data to ensure the model could distinguish between heavy, moderate, and light traffic accurately?

The movement of a vehicle. And the stagnant vehicle, the stagnation where it is available in particular signals how it is waiting. So we will have a three classification, 3 class classifier, multi class classifier. We have used traffic is a heavy or. Very, uh, normal. Are moderate, so we used the color representation. Uh, to carry over the next process or we can do for alternate route in the heavy, whereas in the normal we can, uh, make them to. Uh, work, uh, and then uh. In the moderate level we can wait and then make them to move. So the representations that we have used is. But the color representations we have used the label classifiers are what is. The label classifier we have used is a. ABC classifiers we have used. Uh, for uh, classifying as a three different classifications.

You mentioned using ABC classifiers for multi-class traffic detection—can you explain how you handled false positives or misclassifications during real-world deployment, and what adjustments you made to improve reliability?

False positive or misclassification will affect the. Uh, results, uh, so false positive and misclassifications are very minimum. Uh, in this particular project, uh, where is the false positives, the stagnant vehicles which are not moving, but on the way, the side they have kept on the both sides that is affecting the route that is also classified as a, uh, false class false positive. So we are misunderstanding or assuming, uh. Uh, differently, uh. That it is also coming under the classification. So that is a part sparsity the stagnant makers which is unmoved or UN parked wakers, the parked vehicles, unknown parked vehicles also makes an impact to the false positive stage. So how can you explode is the movement or whether the person is available or it is a.

Right, so when you found that parked vehicles were causing misclassification, did you modify your CNN model architecture, change the input data, or adjust your labeling strategy to reduce these errors?

What's positive and we can remove that for misclassification.

Right, so when you found that parked vehicles were causing misclassification, did you modify your CNN model architecture, change the input data, or adjust your labeling strategy to reduce these errors?

Helping for the misclassification, but we have to remove that. That is not helping. Actually it is a hurdle. So we have to improve the accuracy. Means we can remove these by labeling as a parked vehicle.

Not moving. From the CNN we can in the video. We have received. We can easily classify whether it is a part. Moving vehicle along with the driver.

Thank you for clarifying. Shifting to your teaching approach, can you describe a specific strategy you use to ensure that complex topics in theory or laboratory courses are accessible to students with varying academic backgrounds?

Thank you for clarifying. Shifting to your teaching approach, can you describe a specific strategy you use to ensure that complex topics in theory or laboratory courses are accessible to students with varying backgrounds?