Open Data Science Conference

2019 ODSC Odyssey

PS — This was buried under my draft for so long that I forgot to post it.

If you don’t know what ODSC is, sit down, grab some snacks I’ll walk you through my 3 days journey of ODSC Conference 2019 which was held in Bangalore.

“It is what it is”

ODSC (Open Data Science Conference) is a community-driven organization that aims at getting data-driven people together in an effort to encourage the exchange of innovative ideas and the growth of open-source software.

The conference is all about growing your data skills, the means of which is limitless. You can learn from workshops, conferences, talks and most importantly networking.

Day 0 — Workshop

Kicked off with Registration and Hot Coffee, The registration process was hassle-free, it took less than a minute. Who likes to stand in the registration counter for more than a minute? The answer is none unless you’re the person behind the desk.

Post Registration we entered the Ballroom before the workshop began we had a lite hearty chat with the speaker and all the participants.

The workshop that we decided to attend was about the Recommendation Engine. It was titled “The Art and Science of Building Recommender Systems”.

What is a Recommendation engine?

In simple terms, it’s a system that filters and shows items that the user might be interested in.

Recommendation System can be found everywhere and for many online platforms, the recommendation engine is their bread and butter.

Amazon, Netflix, Pinterest, Youtube, Spotify, and many others built their whole platform around recommendation engines, that’s how they keep users glued to their platforms.

Dr. Sarabjot Singh Anand conducted the workshop and In this workshop, he taught us the algorithms behind recommender systems in different domains and how the domain impacts the approach used.

Types of Recommender systems -

• Content-based Filtering

– Track what the user consumes

– Recommend Items similar to those consumed in the past

– Need item descriptions

• Collaborative Filtering

– Track all user consumption

– No need for item descriptions

– Recommend items consumed by users who have a similar consumption pattern to the user of interest

• Hybrid Approaches

– Combines both approaches to address

The workshop was for 8 hours, it was fine, if only it had a hands-on experience then it would have been great because I learn things by doing it.

The workshop ended with networking.

Day 1 — Core Conference Part 1

Core conference started with a keynote from Dr. Viral Shah (CEO at Julia Computing), in which he talked about Julia and what is their main goal and how far they’ve come. He also talked about their paper which aims to bridge the gap between Scientific Computing and Machine Learning.

He also talked about the Libraries available for Julia, books to refer to, community and also about Juliacon.

He ended his keynote with the quote by Guy L. Steele Jr. “The main goal of designing a language should be to plan for growth. The language must start small, and the language must grow as the set of users grows.”

Post-Keynote, Sheamus McGovern (CEO ODSC) welcomed us with a brief introduction about ODSC and how it started.

We were told to form a group of 10, each being from different backgrounds, to get self-organized. It was a fun activity, I got to know a few people, their motive for attending the conference and what were they working on.

People from 15 countries were attending the event, and there were about 84 speakers from 7 countries.

Post welcome address, the whole area was divided into 5 tracks, 4 for parallel sessions and 1 for networking. We were told about the law of 2 feet, which says “If you are neither learning nor contributing MOVE ON to a place you can”. This is a great move because if you don’t like anything then you can move to another session or you can move to the fifth track for networking and surprisingly it’s the fifth track is where a lot of interesting stories and learning happens.

I’ll try not to talk about each and every session that I attended but, instead of the ones which were really good.

“Making Sense with AI” — Jared P. Lander

Jared is the Chief Data Scientist of Lander Analytics, He is the author of R for Everyone: Advanced Analytics and Graphics, a book about R Programming geared toward Data Scientists and Non-Statisticians. He took a very joyful session in which he talked about what “Statistics” actually is and how buzzwords like “AI”, “ML” have made it more attractive to people and have to broaden the reach.

He defined AI as “What Humans Find Easy but Computers Find Hard”.

Humans find it easy to identify cats and dogs, but Computers find it hard.

“People want to regulate AI but no one cared when it was called logistic regression.”

I had the opportunity to talk with Jared, he extensively works with R and me being a Pythonista, our conversation was quite witty.

Naresh Jain — Ethical AI — Fishbowl

What is Fishbowl?

Fishbowl is an activity, it is similar to a group discussion but less chaotic.

There will be few chairs on the stage, and the host tells the topic and based on the topic anyone from the audience can come up and talk about it if the person has nothing to say he can leave the stage and make space for a new person.

The topic in ODSC was about “Ethical AI” There has been a lot of concerns about the black-box nature of AI. People have been asking for a sensible AI guideline with the weight of Law behind it.

• Transparency: Any time an AI system makes decisions on a user’s behalf, that person should be aware of it. The reasoning behind decisions should be easily explainable.
• Safety: AI systems should be designed to withstand attempted hijacking and other attacks performed by hackers.
• Fairness: Decisions made by AI systems should not be influenced by gender, race or other personal identifiers. They should be as impartial as possible and not reflect human biases.
• Environmental stewardship: Not all the stakeholders in AI development are human. The development of these platforms and the implications of their decision-making and sustainability should take into account the needs of the larger environment and other forms of life.

I really enjoyed this discussion, there were a lot of views from different people both positive and cynical. I leave the link to the video below, take a look at it.

https://www.youtube.com/watch?v=BG-72x-dPvo

Fishbowl was the last session for the day, post that we had dinner and networking.

It was amazing to meet people from different backgrounds and places, we also met a kid in the conference, he was the youngest one in the whole conference, he was just 12 and runs his own company “Power Booster”, with the aim of reducing the carbon footprint on this planet.

I had the best conversation with him, I asked him how did he start to which he replied: “My uncle told me that if I work with python, I’ll make a lot of money, then I started learning python and got to know about TensorFlow, that’s how it all started”. Then I asked him who guides him and his response blew my mind “I guide myself!”. He is homeschooled and he learns all by himself.

I cannot forget this day, this kid inspires me.

And that was the end of day 1, it was a great day from few sessions being great and few were disappointments.

Day 2 — Core Conference Part 2

Day 2 kicked off with Grant Sanderson’s Keynote, Grant is the man behind the famous YouTube channel “3Blue1Brown”. 3b1b centers around presenting math with a visuals-first approach.

Before the keynote began we had a fun little chat with Grant, he told us the reason behind the name of his Youtube Channel “3Blue1Brown”.

“Okay, I’m not going to pretend this isn’t a little weird. I made the logo to be a loose depiction of my right eye color: It has sectoral heterochromia, 3/4 blue 1/4 part brown. It was a way of putting a genetic signature on my work, and the channel is all about seeing math in certain ways. The name, of course, is just derived from the logo. In hindsight, this feels a bit more self-centered than I’d like, but hey, what can you do?”

“Concrete before Abstract”

Grant’s started his talk about adding reactions or characters to their papers or whatever they were presenting, instead of saying plain English “Hey you might not understand that but, in the long run, you will”, you can side square that and show a confused character by this you also add delight and empathy.

He continued with the idea that he keeps in mind while creating videos, i.e

Where do you start? To which level do you abstract so that anyone understands?

He shared this beautiful video of Sample Math Circle Class: Bob Kaplan

https://www.youtube.com/watch?v=tE0k16eAZaA&feature=youtu.be

Put simply, it’s to resist the temptation to open a topic by describing a general result or definition, and instead let examples precede generality. More than that, it’s about finding the type of example which guides the audience to rediscover the general results for themselves.

He ended his keynote with the making of one of his videos about “Neural Networks”.

Post keynote we went on with different sessions some of them worth mentioning would be

• “Image ATM(Automatic Tagging Machine)”
• “Speech Recognition in Bixby”
• “Using Deep-Learning to Accurately Diagnose Your Broadband Connection”
• “Data Science and the art of “Formulation” ”
• “Causal data science: Answering the crucial ‘why’ in your analysis”

Post Conference we had a chat with all the members of ODSC and speakers, we talked about the future of AI, the Community and took a few photos.

Overall the Conference was really good and well organized, though there were a few disappointments I wish they could focus more on the quality of speakers than the quantity but, hey nothing is perfect.