BITS Goa Women in Tech interviewed Aditi Sinha, a 2018 graduate from BITS Pilani, who majored in Economics and Finance. She has since worked with various organizations, including Social Cops, YES FOUNDATION and Hyperloop

In 2019, she co-founded Locale.ai, a start-up which uses the location data companies collect and give them insights to analyze and optimize their ground operations.

How was your life at BITS? What were the obstacles that you faced in college?

At BITS, I decided to not go for any dual degree with my Economics major. As a result, I had a lot of free time at hand that I used to discover my interests. I joined clubs on campus and also took up projects with organizations like J-PAL at MIT, Vision India Foundation. I authored research papers in Economics that got published in international journals. Working with clubs also taught me skills that came handy later. “Sponz” taught me how to convince people to give you money, TEDx taught me how to manage people. My research experience helped me get my first job out of college. My time at BITS shaped me as a person and helped me realize my passion. There were definitely obstacles on the way. My decision to not opt for a dual degree and then later on to not sit for placements wasn’t taken well by friends, teachers as well as my parents. But, I believed in what I was doing and that made all the difference.

How did you decide on a career path?

At the time of my graduation, I knew two things. I was passionate about impact at scale and I loved working on challenging problems in a startup-like environment. So, upon graduation, I took up a job at a data consultancy startup called Social Cops (now Atlan). It was there that I met my now co-founder Rishabh Jain. He had been working on different location projects with governments, FMCGs, and startups. This is when we realized that companies are collecting a huge amount of location data and they just didn’t have the right tools to gain insights from that data. They had to build internal products that were extremely painful to use and couldn’t deliver business insights with the speed that was needed. So, we decided to take the plunge and work on this problem. Both of us left our jobs and moved to Bangalore to build something that would empower local teams to get the right insights from their geolocation data just in a matter of a few clicks

If you could go back and give your college-self a piece of advice, what would that be?

Well, I would push myself out of my comfort zone a little more and try out more new things. Because trust me, you really miss college once you graduate and start working. I would try to fail more and work with more startups in their core teams. I would explore many different things, but would like to narrow down and figure out what I am really good at and go deep into it.

Once you start working, you realize that you have to unlearn a lot of what you picked up during college. Failure is not as big a deal. You learn by doing things and by often failing at them as well. The fact that you can iterate, improve and get better with time is something so fundamental to professional life but our college psyche is modelled in a completely opposite way where every submission you do is final.

To know more about Locale.ai, do check out their website and their blogs.

What is Computer Vision?

Computer Vision is the subfield of artificial intelligence which tries to imitate the human vision capabilities. This not only includes the ability of humans to see, but also the ability of humans to perceive the environment that they see and make decisions based on it. Computer Vision is a broad field, and encompasses not just the machine learning models required to build logical decisions, but also hardware settings required to “view” the environment, and appropriate encoding and signal processing to construct the image to be used by the model as input. The main aim of computer vision is to understand and automate tasks that the human visual system can do. It has applications in several fields, and is widely used for defect detection, Metrology, Facial recognition and in robotics. Due to the vast amount of data that we generate today, Computer Vision has grown by leaps and bounds in recent times.

A ridiculously brief history

In the late 1960s, computer vision began at universities which were pioneering artificial intelligence. It was meant to mimic the human visual system, as a stepping stone to endowing robots with intelligent behaviour. In 1966, it was believed that this could be achieved through a summer project, by attaching a camera to a computer and having it “describe what it saw”.

Studies in the 1970s formed the foundations for many of the computer vision algorithms that exist today, such as extraction of edges from images, labelling of lines, non-polyhedral and polyhedral modelling and representation of objects as interconnections of smaller structures.

The 1980s marked the first time that statistical learning techniques were used in practice to recognise faces in images. Despite the progress, by the 1980s, computer reasoning was still far from achieved. This led to a period of reduced funding and interest in computer vision and artificial intelligence research in general, termed as the “AI Winter”. It wasn’t until AlexNet won the ImageNet challenge in 2002 that AI witnessed the massive boom that it currently enjoys. However, this doesn’t take away from the fact that Computer Vision continues to remain a difficult concept to master, and there still might be a long way to go.

Why is Computer Vision so hard: The Moravec’s paradox

Moravec’s Paradox states that “It is comparatively easy to make computers exhibit adult level performance on intelligence tests, playing checkers or calculating pi to a billion digits, but difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility… The mental abilities of a child that we take for granted — recognising a face, lifting a pencil, or walking across a room — in fact solve some of the hardest engineering problems ever conceived… Encoded in the large, highly evolved sensory and motor portions of the human brain is a billion years of experience about the nature of the world and how to survive in it. ” In short, the hard problems are easy, and the easy problems are hard. This explains why research into computer vision and robotics had made so little progress by the 1970s. Moravec’s Paradox was the reason that a lot of the “AI research” lost its AI label, understandably how can something be termed “intelligent” if it cannot even replicate the behaviour of a one-year old? The invisible complexity of the “simple skills” that we so often take for granted are extremely difficult for computers to master. Since computer vision attempts to proxy as a human equivalent for sight, it has to be able to efficiently replicate all the evolutionary learning that our eyes and brains inherit for the purpose of ‘seeing’. Computer vision, therefore, always has an uphill battle in front of it.

To put it succinctly, challenges in computer vision are:

1. Computers do not have built-in mechanisms to extract higher-level information from images like humans do.
2. There exists an issue of data representation.

Applications of Computer Vision

• Object Classification
• Object Identification
• Object Verification
• Object Detection
• Object Landmark Detection
• Object Segmentation
• Object Recognition

The backbone of Computer Vision: Convolutional Neural Networks

To overcome the challenges faced in computer vision, feedforward neural networks just won’t cut it. Feedforward neural networks are fully connected networks, and for large input vectors the number of learnable parameters is magnanimous. This leads to a large training time and also makes the model prone to overfitting on the training dataset.

As an example, look at the feedforward neural network above. All the 10 inputs of the input layer are connected to all the 8 neurons of the hidden layer. That is 10*8–80 parameters — just for the first hidden layer! In total, this extremely small feedforward neural network model has around 6000 learnable parameters! Now imagine if we had to use a feedforward neural network on an image dataset. Even the smallest possible image size is 28x28 or 30x30. This requires a feedforward neural network with 784 (or 900) input neurons in the input layer. As the image size increases and the model becomes more complex, the number of learnable parameters for a feedforward neural network reaches to the order of millions!

This problem begs the question, can we have Deep Neural Networks which are complex yet have fewer parameters?

The convolution operation

Our challenge at hand can be solved beautifully by a very simple mathematical tool called the convolution operation. Consider pizza waiting times at a busy Italian restaurant.

At t0, the waiting time for the pizza is X0.

At t1, the waiting time for the pizza is X1.

At t2, the waiting time for the pizza is X2.

Now, if you were asked to estimate the waiting time for your pizza which you ordered at t3, you would probably average all the neighbouring waiting times out.

Estimated waiting time X3 at t3 = ⅓ * (X0 + X1 + X2)

Now suppose, you want to introduce a prior belief to your estimation that the neighbour closer to t3 will have greater influence on the estimated waiting time for t3 than the neighbours before it, you could come up with a increasing weight system W0<W1<W2 such that:

Weighted estimate waiting time = W0*X0 + W1*X1 + W2*X2.

In the above equation, you have basically assumed that information gained from the closest neighbour is higher than the neighbour which is farther away. A different setting of weights could have been done depending on what the information gained from neighbouring cells in the given setting was.

What we have done above is known as a convolution. A convolution is the process of calculating the weighted average of all the previous neighbours to estimate the value of the current neighbour. A convolution is applied when information gain is achieved not just from the current point of interest, but also from the neighbouring points.

1D convolution

Below table is an example of a 1D convolution.

We say that the W vector convolves over the X vector and the result is 1.80 (which is summation of Wi*Xi)

2D Convolution

In a 2D convolution, the information gain is across neighbours in both rows as well as columns. A 2D convolution operation is pretty useful in the context of computer vision, since images are typically represented as matrices. The 2D convolution is a useful tool to analyse images, since a pixel by itself does not give any information gain. We need to look at neighbours along both rows and columns to get some information. Blue coloured pixels could be a part of the sea or the sky, and we can only be sure of which category they belong to once we have looked at the neighbouring pixels. Thus, the convolution operation becomes the foundation for our computer vision problem. We call the image matrix as the input, and the weight matrix as the kernel or the filter. The kernel is used for specific feature extraction depending on what the weights are.

The Convolutional Neural Network

We can now use the convolution operation to build our network. The filters become analogous to weights in the feedforward neural network and the entire convolutional output corresponds to the whole layer of neurons in the hidden layer. In FNNs, we consider all input values from the previous layer multiplied by their weights while in CNNs, we consider only a small number of input values multiplied by the filter values. Thus, CNNs bring about a huge reduction in the number of parameters by making use of two very important concepts — Sparse connectivity and Weight sharing. Sparse connectivity refers to the reduced number of connections between two consecutive layers, while weight sharing refers to the reduction in the number of parameters caused due to the same filter convolving over the entire image. The end result is a neural network which is much more tuned to handling images and is one of the most widely used neural networks for computer vision problems.

Additional Reading

• Intuitively Understanding Convolutions for Deep Learning
• Stanford Spring 2020: Convolutional Neural Networks (CNNs / ConvNets)
• Feature Visualisation
• Convolutional Neural Networks (LeNet)
• Transfer Learning with Convolutional Neural Networks in PyTorch

About the author: Arshika Lalan

I am a pre-final year Msc. Economics and B.E Computer Science student. I am a Deep Learning and Machine Learning enthusiast, who loves to experiment and explore new ideas. I love maths, econometrics and statistics. I also love to work on web development using React.js. I also enjoy writing, and write poetry and tech articles in my free time. You can stay connected with me via LinkedIn and check out my articles on Medium.

Thank you for reading!

Check out more from BITS Goa Women in Tech on our Website , Instagram, LinkedIn and Medium!

Hello reader!

I am Ayushi Dubal, and I am a computer science student at BITS Pilani, Goa Campus. Recently, I received the Google Women Techmakers Scholarship for the APAC region. This article is to provide some insight into the application process. I hope you find this helpful!

What is the WTM Scholarship Program?

Dr. Anita Borg is one of the world’s most respected computer scientists, who strove for technical fields to be places where women were equally represented, places where women could contribute to, and benefit from technology. In honour of Dr Borg’s lifelong efforts to break down all barriers that kept women and minorities from doing so, the Google Anita Borg Memorial Scholarship Program was founded, which has since been renamed to the Google Women Techmakers Scholarship Program. The scholarship hopes to inspire and motivate countless women to become active leaders and creators of technology. This year, 74 scholars were chosen from across India and Greater China.

Eligibility

To be eligible to apply, applicants had to-

• Identify as female.
• Currently be enrolled at an accredited university for the 2020–2021 academic year.
• Be in year 1 or year 2 of their undergraduate program (new change this year)
• Be studying computer science, computer engineering, or a closely related technical field.
• Demonstrate a good academic record.

The Process

There were three rounds to clear-

• Question round
• Google Online Challenge- Technical round
• Telephonic conversation

Question round

Here’s what the application consisted of-

I. Answers to two essay type questions

These questions could have been answered in either a 2 minute long video or a 300–400 word essay (whichever the applicant was more comfortable with), to assess academic excellence, leadership qualities and passion for computer science.

I chose to answer as a written essay.

The questions that were asked were-

1.What computing technology are you excited about, and how is it changing your region?

2. a. Describe a project (or projects) where you have addressed problems caused by the lack of diversity and inclusion in computing. Consider how many people you may have reached, and the impact of your work on others.

2. b. What problems are caused by a lack of diversity and inclusion in computing? How can these problems be addressed?

There was a choice in question 2, either one of a or b could be answered. I answered b.

II. An up-to-date CV

For the written essay questions, I would suggest that you answer the questions with concrete reasons and personal experience. For questions like the first one, make sure your answer is technically heavy, but try to avoid jargon. Keep your language simple, and make sure you format the document well before submitting it as a PDF.

Try to back your CV up with links to any projects you may have done.

Most importantly, get your application reviewed by at least two people- they will provide you with invaluable feedback.

Google Online Challenge- Technical Round

The next step was the Google Online Challenge, which was the technical round. This round was not that easy, as we had to answer 30 questions in 30 minutes. The challenged comprised of questions on fundamental areas of computing, C programming, OOP & C++, and data structures & algorithms. There was no negative marking in this round. A good foundation in basics of DSA and OOP was needed to solve these questions.

Telephonic conversation

After about three weeks, based on the results of the previous two rounds, I got an email saying that I had been shortlisted for the telephonic conversation round. It is generally a 30 minute phone call with a specialist from the APAC WTM team. The phone call is just a conversation to learn more about your goals and interest in the scholarship. Questions were asked to gauge interest in the field, leadership ability and engagement with the community. You can also expect to be asked to elaborate on a project you have mentioned in your CV.

To prepare for this round, I would recommend going through the answers you had submitted in round one very well, so you have a central idea to talk about while answering questions. Go through different articles and gather your thoughts by writing them down a few days before. This way, you will have a clear idea of what to say when a question is asked. Of course, the interviewer is very friendly and there is no pressure at all, so you can be at ease. My interview lasted just fifteen minutes, but it was a pleasant experience. Also, be sure to ask your interviewer a question at the end- they are Googlers, and you probably have a few things you want to ask them, so make the most of the opportunity!

Retreat

After three more excruciating weeks, the results were declared. I received a congratulatory email telling me that I had cleared all the rounds, and that I would be receiving a 1000 USD cash scholarship and an invite to the scholars’ retreat. The scholars’ retreat is generally an all-expenses-paid trip to a Google office somewhere in the APAC region, but it was held virtually this year due to COVID-19. The retreat was an excellent opportunity to connect with fellow scholars and Googlers, who give various speeches and conduct different workshops. These sessions were aimed at personal and professional growth and were a great experience. Also, you get a bunch of swag from Google!

The Google Women Techmakers Scholarship Program is an excellent initiative that encourages women to look at computing as a career choice more seriously, and I encourage each of you to apply for it. Every step of the way, when you look for answers to the questions asked, you learn so much about the world and yourself. That, in itself, is such an enriching experience that it cannot be missed. I hope you get to experience the journey that I have via this program.

If you have any questions regarding the scholarship and the application process, please feel free to reach out to me at ayushidubal5@gmail.com, and I’ll try my best to help you out :)

Looking forward to the future!

All the best!

Ayushi ❤

For getting to know more about our initiatives, check out our website: https://bitsgoawomenintech.wixsite.com/bgwit

Though BITS always provided us with the same platforms and made no distinctions based on our gender, we needed to work harder to do as well and search longer to find the same opportunities, simply because we were not a part of the network through which our male classmates shared information and support. While applying for an opportunity, we struggled to find peers and seniors who were doing or had done similar things and could provide us with feedback, guidance or even encouragement. After speaking to several female alumni as well as juniors who had all had similar experiences, we realised that something needed to be done.

We, the women of BITS Goa were in pressing need of a community. A space where we could share resources, offer support and guidance, and discuss ideas. A network through which we could collaborate, and work together. This is why we created BITS Goa Women in Tech.

We were all in our 4th year of college, and the need for this community really resonated with us. Although it was overwhelming initially to start something from scratch, especially remotely, the journey so far has been extremely fulfilling. Watching the BGWiT family grow, and receiving messages from our juniors about how much this community has helped them, has really made all of our effort and work worth it. As an added bonus, the experience of starting this initiative has strengthened our friendship, something that we are all very grateful for.

We try to work as openly and transparently as possible, take constant feedback and encourage interaction. We also ensure that all of the material we use and share is stored in a public repository (our website) so it can help anyone at any time.

We really hope that BITS Goa Women in Tech can help you in some way or the other!

For getting to know more about our initiatives, check out our website: https://bitsgoawomenintech.wixsite.com/bgwit