Good day, G. Anitha. I'm Arjun. I've had a look through your background and your professional record. I'm interested in hearing more about your work and how you see yourself fitting in here with our faculty. Ready to start?

Yes. Good afternoon. I'm ready to start.

Good afternoon. Please start by introducing your professional journey in academia.

Your voice is breaking. I could not hear your sentence fully.

Apologies for that. Could you please start by introducing your professional journey in academia?

Yes, good afternoon myself, Anita Gopalakrishnan. I am working as Assistant Professor in the Department of Electronics and Communication. I have recently finished my PhD and out of the PhD I've got to say papers one got published and indexed in biomedical signal Processing and control and one another paper which got published and indexed in the Knowledge and Information Systems. Both of them belong to the Q1 and Q2 journalists respectively.

So this is my profile and I'm interested in the research and my research domain is machine learning.

Thank you for sharing your academic journey, Professor Gopalakrishnan. That's valuable to know.

Professor Anitha, I will now assess your competencies in alignment with the professorial role we are hiring for. Let us begin. First, could you elaborate on your experience and technical expertise in Image Processing?

Yeah, image processing. I have undergone A faculty development program and conducted by the SRM which is actually an AACT sponsored 1 and there I have learnt about the new skill which is QIS. From the QJS. I am able to understand how the simulation happens and how I can map the certain things like water reservoir and a few other soil erosion and things like that inside the map. So QJ.

Is the new tool that I have learned right now and right now I'm just trying to explore the ARC GIS.

Yeah.

Yeah. Basically in order to understand how the, for an example, let us take an instance like water reservoir availability. So across the India, we wanted to understand where all the data of water, I mean water reservoirs are available. So for that purpose what we can do is we can get the data set and from the data set we can map different locations inside the data and just apart from this.

Take your time. How do QGIS and ArcGIS enhance your ability to process and analyze image data effectively?

Image processing, my work is basically my PhD work is basically just to develop a model which will understand to classify the different levels of anxiety. So for this anxiety level classification, one important parameter that we can utilize is nothing but the facial recognition. So in the domain of facial recognition also this image processing leaves a major role in that. So I would say this image processing is a major thing that I have also adapted in.

I have a very good connect with the.

My work.

Yes.

To clarify further, could you elaborate on the specific image processing techniques or algorithms you employed in your PhD work related to anxiety level classification?

Yeah, actually right now what I have done is in my physiological.

Data collection. I have incorporated few machine learning algorithms, the first one being the SVM, Support Vector machine, then KNN.

And quadrature discriminant analysis and Gaussian process and things like that. So out of that, this is one set of understanding things from the physiological data, various physiological data that I have taken up as nothing but the galvanic skin response, then respiration, then temperature, then saturation.

I mean oxygen saturation, perfusion index and things like that one, another domain that I can explore as nothing but the facial recognition. So facial recognition is the one that I'm right now exploring. So for that I have tried doing it with the deep learning. For the deep learning concepts I've learned about the CNN. So with that LSTM and CNN I'm trying to deploy.

Understood. You’ve applied a broad range of techniques—from machine learning models like SVM and KNN to deep learning methods such as CNN and LSTM—for both physiological data and facial recognition. This demonstrates a substantial understanding of integrating image processing and advanced computational techniques in your research. Let us move forward. Could you now outline your expertise in Embedded Systems and Communication, and any relevant applications or projects you've worked on in this domain?

Are you telling me to if I mean speak about the PST domain that I have explored?

Yeah.

Yeah, right. Yeah, right now the first and foremost thing that I have explored is nothing but identifying and classifying the level, severity level of the anxiety. So it could be anyone, OK, It could be you or me or anyone. So everyone will have a cognitive load on their mind whenever a task has been given to them at the real time circumstances. It is mandatory to understand how much a person is undergoing A cognitive load. So in order to understand how much.

A percentage of cognitive load is able to be bad. We wanted to understand what is the threshold level that we can keep that will just classify the level of anxiety from being more severe and the bearable 1. So these two binary classifications are being done in my project and this is 1 good application that I would say. So this project I just wanted to expand using the. Most important application which is nothing but identifying the anxiety level of.

Take your

Artisans, children. So I'm in the process of collecting the data of the autism kids where I'll be able to identify the level of severity in them, and this will be very much useful for the healthcare caretakers and the practice nurse.

Thank you for sharing, Professor. While your work in anxiety level classification is remarkable, it seems more aligned with your research in physiological and machine learning applications. Could you now provide details specifically about your experience or projects involving Embedded Systems and Communication, particularly any technical implementation or hardware-related expertise?

OK, so the very first challenge that I had in this particular research is nothing but we were hunting for a very good data set which is reliable, which is having a very good outreach and it should be offer non invasive method. Most of the research that we see in the previous in the literature is nothing but more of invasive technologies that includes you know like ECGEMG and. I mean, yeah, EEG and things like that.

Which are actually trying to stimulate, uh, more anxiety to the people. So what we decided is let us not just rely on a, a different data set. Why don't we come up with ourself a data set which can, uh, introduce, I mean, which can, uh, measure different physiological signals like galvanic skin response, then heart rate variability, perfusion index and saturation of oxygen. So these are the physiological parameters that have been recorded and we have got.

Data set which we called as a physiosense data set which is exclusively been collected by us and our team. And this is being actually taken the help from the human physical committee. As we all know anyone who is going to work with a human involved inside the research have to get the ethical committee. I mean, I mean they have to get the ethical committee allowing. So we have got the ethical committee.

People allowing this and then they have guided us how to take care of the data set collection and we have done that successfully and we have come up a data set which has been coined with the name of Physiosense.

Understood, Professor. Your development of the Physiosense dataset, which is based on non-invasive physiological measurements, highlights your comprehensive approach in tackling challenges. However, I noticed the focus has been on data acquisition and processing rather than embedded systems directly. Could you elaborate on any specific embedded system development or implementation you've contributed to, such as hardware interfacing, microcontrollers, or firmware used in collecting this data or other projects?

Basically, I have used the Ordino Ordino Uno hardware kit and for that Ordino you are no hardware kit. We have embedded 4 new sensors. As I told you before, it was a galvanic skin response sensor which will try to investigate the electrodermal activity. One another sensor is the heart rate sensor and we have also embodied the oxygen saturation sensor along with the perfusion index sensor. So in this journey, I.

Have undergone, I mean we have come up with an hardware which is really able to capture these physiological data. So for this we have the I mean this particular kit has got such good number of parameters capturing sensors. So this is the one that I have experienced I would say in the terms of embedded.

Yeah.

Thank you, Professor. Your integration of Arduino Uno with multiple sensors like galvanic skin response, heart rate, oxygen saturation, and perfusion index showcases your hands-on experience in embedded systems with a clear application to physiological data collection. Now, let's proceed. Could you elaborate on your approach to teaching both theoretical and laboratory courses effectively?

So I have 13 years of experience in electronics and communication engineering. I'm gonna taking the core subjects. I have handled communication theory, then digital waveform coding and communication techniques, modulation techniques and then I have handled also the ad hoc and wireless sensor networks. Also I'm good at taking the PG level courses such as machine learning techniques. And one more paper that I would say that I'm good at doing is nothing but data analytics with Python. So along with the classroom.

Study, I always believe to show and visualize the students, how the things are happening. And I would say I'm good at structured way of communicating the concepts. So this way I would say I'm these are the papers that I have taken. Anything beyond this, any paper has been given to me. I would surely show my interest in improving my knowledge and then disseminating the information to the students. And apart from this laboratory courses, I have handled the circuits and simulation laboratory.

Different rectifiers are being handled and then different methodologies like simulation, MATLAB are also being handled. So apart from this, I have also handled a digital electronics laboratory where half subtractor, full subtractor and things have also been done. And also I have a good working level in the Xilinx software and I'm also good at working in Python collab, Google collab, Python notebook.

Thank you, Professor. Your comprehensive experience in both theoretical and laboratory courses across a range of subjects, along with your emphasis on structured communication and hands-on visualization for students, is clear. One follow-up: could you describe a specific instance where you tailored a complex theoretical topic to make it more accessible for your students?

OK. I would say any student who comes inside the electronics and communication department, it is very mandatory for him to understand the concepts of communication basically. So a second year student maybe if you could take that person should be good at understanding what is a modulation concept is all about. So in order to tell him what is a modulation concept, I would say or modulation is of course we will say the definition, the process of changing the characteristics of work.

Carrier signal in accordance with thought of the parameters of a modulating signal. So amplitude. So this is actually the basic definition of a a modulation. So for that what I usually do to my students is I would take an example like a small kid is there that small kid we have to depict as our low frequency message signal or a modulating signal and that low frequency modulating signal cannot be just sent as such because it.

Of low frequency and there are more options of getting lost in the environment. So for that reason we would need a carrier who is to be depicted as a like a elderly person or a caretaker who will be carrying the kid from the school to the house and house to the kid. So the similar way we need a high frequency carrier signal to carry the information from the message from the sender to the receiver. So this is how we will be able to.

To tell the students like any concept, any concept that you take will be able to explain them as if we can explain to the five years old kid. And this will be helpful for the, you know, students to understand in a better way. So the modulation concepts are being summarized very similar to a kid being carried from school to the house and house to the school. So this is one such example of of how I can jot down.

The complex, uh topic in tourism.

Thank you, Professor. That's an excellent analogy, using the concept of a child and a caretaker to illustrate modulation. It effectively simplifies a complex topic, making it relatable and accessible for students. Now moving on. Can you outline how you approach student evaluation and assess their performance consistently and fairly?

OK, Uh, basically as a teacher, we cannot, uh, just consider all the students to be of the same standard. There's going to be a diverse, uh, students level 1 being little poor and studying. Maybe you would say a weak learner, one maybe at a high level of understanding how much ever you give or whatever that you give, they will grab. So we should have the patience to tell them how things are working. So you cannot just disseminate the information as such. So you will have to customize the.

That you disseminate the information to each and every person and also I would do the evaluation in such a way that of course depending on their level of understanding only.

Like for a week learner, we must give away certain problems, uh, so that they'll be able to understand the concepts and they'll be inspired, inspired and learning further. And we will have to give little challenging problems to the high level students so that they feel that a little more, you know, informations have to be learned and how this can be taken up for a research level or something. So involving basically this UG students.

I I'm, I'm I'm totally into it.

For the research level.

Sometimes application.

Thank you, Professor. Your approach to tailoring evaluation methods to match the varying levels of student understanding, while challenging high-level students and encouraging research involvement, is commendable. Could you share a specific strategy or tool that you use for maintaining objectivity and fairness in assessments?

Yeah, uh, assessments I would date, uh, I would say, uh, any theoretical concept that we do. For an example, once I remember I have taught about the double sideband suppressed carrier system to my communication students. So to them what I did is nothing, but they, I asked them to come up with a program, maybe using a Python to give me the code level.

So that to understand the concepts of double sideband suppressed carrier, I never told them.

You know, brought, I mean, I never told them what to be done, but I told them this is the one that I'm expecting, just the output. So different members, I mean different students tried their level and they came up. And for them, I have assessed them and I've given the grace marks like every week, once every theoretical topic will be understood along with the simulation, digital simulations. So by this way, they'll be able to come up with certain outputs. So depending on the output.

And their outcome, level of outcome, I'll be giving the marks to them.

Yeah, basically we should try to understand what is their interest first of all, rather than just enforcing them what we need. It's better we understand their domain of interest. Whoever has been, I mean, showing that interest in the domain which is in close aligned to us, we can take up fully and then, you know, we can guide them. First of all, I would ask them to go with the literature survey, leave the literature service quite, you know, like pleasing.

Understood, Professor. Your strategy of integrating practical simulations, such as asking students to design Python programs for double sideband suppressed carrier systems, is an excellent way to assess their theoretical understanding and their application skills objectively. Incorporating such creative problem-solving tasks indeed inspires deeper engagement and learning. Let us proceed. How do you mentor students for their projects and research work effectively?

And if they have done enough of literature survey, in that case I will, I will ask them to come up with a digital simulation. And that digital simulation, if they have done well and if they are able to just interact with me and ask if any, I mean, if any queries are there, I'll be there to help them at any cost, even at the hardware level and the simulation level. So once they are done with the simulation, I would definitely increase them to go with the hardware. And if the hardware works, we will together.

As the team will work for the funded project, maybe for the different government opportunities like DST and CERT. So one such thing that what we have done is and the collection of my Physiosense data set. As I told you before, I have involved my UG students in developing the hardware kit. So right from collecting the sensors and then embodying them into the ordinal board. Everything from the motivation of the teachers and our team.

We have made the students do it entirely.

Yeah. So the method that I would take is nothing but everything when it has been shown as a visualization, but replacing visualizations and simulations that would reach the students very well. Even if it is a very chaotic topic, let it be a very confusing topic or something like very complex one. It is always we have to think that we are disseminating the information to a small kid in such way. We will have to do so once it has been done with the proper visualization and.

Thank you, Professor. Your hands-on and student-centric mentorship approach—starting from understanding their interests, guiding them through literature surveys, and progressing to simulations, hardware development, and even sponsored projects—is an exemplary model of effective academic guidance. Utilizing projects like the Physiosense dataset to involve students in real-world research and hardware development reflects a strong commitment to both teaching and innovation. Next, let’s discuss your communication style. How do you ensure clear and structured delivery of complex ideas during your teaching or research presentations?

Simulation, definitely it will reach the students and of course a real time examples are very much important for them to be given. So when they try to connect this particular theoretical topic with the real time application, it will definitely help them to understand the concept very well.

Thank you, Professor. Your emphasis on visualization, simulation, and providing real-world examples to make complex topics accessible highlights a well-structured and effective approach to communication. It not only simplifies learning but also bridges theory with practical application. Now, let us discuss your research background. Could you provide insights into your publication experience in reputed journals and the focus areas of the works you've published?

Yes, sure. So my PhD work has given two different publications. 1 publication has happened in biomedical signal processing and control which is an elsewhere publication and that has been told as AQ1 publication journal and with the impact factor of 4.7 which is of high impact factor. And my second paper got published and indexed in the knowledge and information system that.

By the Springer publication which has nothing but a Q2 journal that has also been recognized and the well reputed journal. So these two are my publications. They have a very good impact factor, the first one being 4.7, the other one being 3.5.

And other than that, I have also close to six to seven conference papers that I have collaborated with the UG students and few with my colleagues and I have come up with seven close to 7.

Conference Papers.

Thank

Thank you, Professor. Your research contributions, including high-impact journal publications in Elsevier and Springer, along with collaborative conference papers, reflect a strong commitment to advancing and disseminating knowledge in your field. Lastly, could you outline any experience you have with industry projects or consultancy work, if applicable?

Yes. So right now I have been involved in a project that has been given by the techno notes of IIT column.

So that industry has demanded a project which I'm right now doing. This is actually little extension to my PhD work. Now what I'm doing is I'm extending this to the quantum machine learning intelligence with the transfer learning involved in it for a physiological and health analysis. So this is a thing that I have started to learn and I have started to work on it. So this project is actually being done for the Technon.

Kerala.

Thank you, Professor. Your engagement with Technonotes from IIT Kerala, particularly in extending your PhD work into quantum machine learning and transfer learning for physiological and health analysis, demonstrates your commitment to applying your expertise in cutting-edge industry projects. With that, I have gathered the necessary information to assess your expertise and suitability for the professorial role. Allow me a moment.

To begin, Professor, could you share an example of a foundational course in your field that you find particularly rewarding to teach, and why?

Will be benefiting the students to the fullest I believe.

Thank you, Professor. Teaching communication theory and integrating machine learning into your courses indeed sounds impactful, especially given your research expertise in the area. Let’s move forward.

Professor Anitha, your research background includes some notable projects, such as using machine learning to develop a non-invasive method for determining total body water, and creating an RFID-based detection system for locating misplaced objects. Let's address the first project on determining total body water. Could you explain how you employed machine learning in this context and which algorithms or methods proved most effective?

Yeah. So first of all, I would say basically the total body water management system, we developed a hardware where it has, it has, I mean there are four different levels of sensors have been involved in that. So one of that sensor is nothing but the water level sensor and mineral sensors are also been involved in that. So these particular sensors have been imported into the Ordino board, so and this particular.

Sensor has, I mean tried capturing the data from different person where the Velcro, you know, probes are being attached to the person's fingers and for a certain period of time they have been left as such without doing anything. The physiotherapy data has been recorded from that and from that particular understanding from the literature we have assessed how much is the level of water a person has got in his body. So different algorithm that we have involved.

K nearest neighbor algorithm has also been involved in These are just the two algorithms that I'm just getting right now. But beyond this 2-3 algorithms are also being done like Gaussian process, then quadratic discriminant analysis and like you know like a few other like you know nail base are also been involved in this level.

Something but SVM, support vector machine and then KNN that is.

Understood. It seems you've integrated various machine learning algorithms like SVM, KNN, Gaussian process, and others to analyze data from sensors and physiotherapy metrics to estimate total body water. A quick follow-up: How did you determine the most suitable algorithm for this task, and what metrics or criteria did you use to compare their effectiveness?

Your voice was breaking, can you just repeat it again if you don't mind?

Yeah.

Yeah.

Go ahead, Professor. Take your time.

Sure. I have assessed with few parameters performance metrics like accuracy and then F1 score, recall, precision. So these are the major parameters that I have assessed that from the level of accuracy and precision, whichever the algorithm is performing well, we have understood that that particular algorithm is performing well and then we have tried making it as an like an ensemble. Basically the ensemble is an extended idea.

From that paper, but still we have tried with all this performance metrics as I told you before, accuracy and then F1 score, precision and recall.

So out of that SVM was performing well.

Thank you for that clarification, Professor Anitha. It's insightful to know that SVM emerged as the most effective technique based on metrics like accuracy, F1 score, precision, and recall. Now, moving from your research to teaching—how would you simplify and explain the application of such machine learning algorithms, like SVM, to students who are new to this subject?

OK, definitely. So I would start with a simple binary classification where would we would classify into two levels, let the levels be 0 and the other level to be one. OK For an example, we will take an application, we wanted to assess whether a person has has got you know a malicious note, I mean has got a cancer note or not. So we wanted to assess whether a person is cancer affected or not if a person is.

Is it affected? We would say that to be 1 and if not, it to be 0. So for that we will have to, you know, accumulate different data points. Those data points are been put on the space and we will have to check whether that particular data is a linear method or not. If it is linear, it is quite obvious that we can separate them into two points. So there's a conference called as a hyperlane in support vector machine. So what does the hyperlane do? It will typically work like a margin.