Skip to content

R3plica

AI Is Like Teaching A Baby To Understand Things

Deep Learning, AI, Computer Vision, Blockchain7 min read

Mahender Reddy

Meet Mahender Reddy

Deep Learning Researcher @ Adventum

Mumbai, Maharashtra, India 

Mahender started working with Python in 2017 and in the process completed his graduation at BML Munjal University.

He started working on machine learning in 2018 and completed many projects in Deep Learning such as t-shirt color detection, spam classifier, prediction of network attacks and smart text recommendations as well.

At the moment he has been working in AI from the past two years and started working as a Computer Vision and Deep Learning research intern in Adventum


When did you first hear about Artificial Intelligence?

I heard about AI in the year 2017 and on Friday at that time I was in my second year of BTech and one of my friends (Battu Sai Krishna) was studying something in the library and I went near him to tease him since we usually don't go to the library.

When I peeped into his laptop he was studying something related to Logistic Regression.

I asked him whether this is part of our syllabus then he said no it isn't. I am just learning something from outside.

I enquired him what he is studying and all. Then he started explaining to me about AI and its Applications.

Later that day I went to my room and started researching it.

How AI is an emerging subject that will revolutionize the world?

This made me more curious and then I asked him how to start and what is available.

After a long discussion, he suggested that I start with the Applied AI Course and also contact another friend who was already into it (Maheswar Kuchana).

I asked Maheswar what are the prerequisites and how to start? and where to start?

Then After that I started learning python and in the meantime, I was brushing up my statistics skills and then after the completion of the python programming I started my first course of Machine Learning (Andrew Ng) parallelly with Applied AI Course and it is a wonderful course to start with to get an idea of AI, ML, and DL.

We all Know about AI from the long back it is used in many movies such as James Bond, Terminator, iRobot and Matrix (My Favourite Movie) but recently people started focusing more into this direction and implement these things even though AI has been there for a long time but recently came back to life due to High computation power and large data (It's always hungry for data).

Define AI and what could be the impact on society?

Artificial Intelligence in simple terms is like teaching a baby to understand things once it understands, it will do everything by itself.

A Normal program is all about creating the rules and boundaries to get the results whereas in AI it will make rules by itself to map the input to output.

AI will create a huge impact on society. It is going to make life easier and also better at the same time.

Right now we all are in crisis due to COVID-19 at this time too we can have an AI application to help us in driving cars (Self Driving Cars), Delivering the goods (Drones with object detection), Screening the People, Entertainment (Netflix, Prime) Recommendation systems.

AI can be used for good as well as bad purposes. With the help of AI, we can create a Killing Drone Machine and it will be too difficult to control them as well.

Joseph Redmon, the creator of the popular object detection algorithm YOLO (You Only Look Once),

Tweeted on 24th Feb 2020 that he had ceased his computer vision research to avoid enabling potential misuse of the technology and his research for the Military purposes and privacy concerns.

Elon Musk (CEO of SpaceX and founder of Tesla) called AI is humanity’s “biggest existential threat” and compared it to “summoning the demon.”

Using AI for the Bad purposes can lead to Destruction of the Humanity but if it is used for the good purpose it has already made a breakthrough in all the industries mainly (Healthcare, Automobile) so AI can make good as well as the bad impact on the society based on the human who is creating it.

It all depends on us how we want AI to make an impact on the Society.

How did you become an AI researcher?

Self-Learning, College Projects, Freelancing. As I completed my Machine Learning Course (Andrew NG) I started reading articles on Machine Learning and AI all the time in Medium.

I used to get the notification every morning and I used to complete them one by one in my classroom.

This was continued every day and then I started working on some projects with Kaggle Dataset. I started with the Iris dataset, MNist dataset, etc.

As I was moving forward I started working more on the projects, even our teachers used to give projects in Machine Learning and Deep Learning. This way I started gaining knowledge on AI.

I Used to download many datasets in my hostel since I didn't have WIFI at home. During Holidays I used to apply all the knowledge I had on that datasets and explore them.

As my curiosity increased day by day, I started learning about new algorithms exploring deep into AI.

Later I got some projects from friends (College Assignments) and my connections too who wanted some help in building AI Applications so while helping I was also learning and gradually became an AI researcher.

I always wanted to try out new things and was very much interested in technology and their innovations.

This made me think of applying AI in different technologies and then I used to search for those technologies.

If they are already present I will see how they are implementing the things. Thus I became a researcher and then shifted my interest into AI becoming an AI researcher.

What is your main interest in working on deep learning and computer vision?

When I started working in AI Domain, I started learning Machine Learning and then slowly dived into deep learning but when you are working in Machine Learning we need to pre-process the data to make the data suitable for our model to train and the best thing about the Deep Learning is that it mimics our brain (works just like our brain does) and it defines the features by itself that define the objects in our model.

We can think of it as a baby (DL Model) we just need to provide food (data) for it to grow (predict).

That's what took my interest and DL is more about matrix multiplications and additions (more of transferring signals from one place to another).

Convolutional neural network (CNN), Long short-term memory (LSTM) took my mind away from that each of them accomplishes the things in their own way one takes and maintains the information from the images and other stores the information from past (memorization) Both these techniques has taken Deep Learning to another level and made dreams become reality.

GAN's are the best algorithm I have ever seen and its hybrids have created Real Human Faces (Not Real People Though!) and style transfer etc.

It has blown my mind and that's where I started focusing on Deep Learning More.

Making Autonomous Cars, Drones, Trucks, etc. was one of my dreams. To be more precise, to create a car that doesn't need any person's support to drive.

When I started working on AI this was one the use cases I have come across that drove me to pursue AI.

When I started researching how to make autonomous cars I came across computer vision which deals with the images and much more (whatever we can see).

Then I started working on Images using CNN but then there came the different filters and techniques to extract the features from the images.

I couldn't understand them well in the beginning then I opted for the Digital Image Processing (DIP) course in my University and my mentor (Devanjali Ma'am) taught us really well and explained things really well about all the techniques and also their needs.

This way I have come into the computer vision field and right now I am working as a Computer Vision and Deep Learning Researcher as well.

Currently, I am working on how we can make our model doesn't get affected by the Adversarial input. There are many techniques to reduce the adversarial but still no solution to eradicate it.

I am working on a solution that includes Blockchain and AI to build a secure and immutable Model that can be trained very easily and increases transparency (Main Concern of Not Using AI).

What goal you would like to reach with AI and Deep Learning?

AI and Deep Learning can be used in Both good and bad ways but it always depends on us how to use it.

With the help of AI, we can create a world with zero mistake death cases in healthcare. I believe we will see in the future there will be an AI assistant to every doctor to make surgery and predict the disease.

AI system to detect the symptoms in people before anyone knows about it. Detecting the disease with more accurate results.

AI is helping hands to take care of Old age People. Autonomous Vehicles and Drones to deliver the goods to your home.

A Personal assistant that takes care of your tasks and meeting etc. A Mixed Reality Games (Sword art Online), AI will be able to get the most of everything.

Using AI in the education field to create 3d examples of all the contents and generating the content too.

My goal is to build an AI system that can be used in every person's smartphone that helps them in their depression time and be there for them when there is no one with you.

This smartphone will be completely based on AI that will take decisions based on your actions but never harms you and this smartphone uses blockchain technology to provide security and transparency.

I know we still have a long way to build such a system but this dream makes me move forward and do hard work.

Building a completely autonomous car is not a fantasy anymore since Tesla already made semi-autonomous cars, so within two-three years we will be seeing the complete autonomous cars on the run.

it all depends on people how they take it whether they want to adapt to this upcoming awesomeness or just stay the way they are.

Whatever it is AI is going to change the future and the way we live, we need to just keep on rolling forward with it.