GIVEAWAY INFORMATION:  

Thanks to O'Reilly and the authors, we are giving away 5 copies of the Practical Natural Language Processing book.   

Giveaway tweet: [TBD at 7:30PM IST]  

To participate in the giveaway, retweet and comment about your favourite part of the conversation with the #practicalnlp hashtag. The winners will be selected and notified on October 1, 2020. To be updated about the results, subscribe to the Youtube channel and follow me on twitter https://twitter.com/mkulkhanna.  

You can also get a 30-day free trial from the O'Reilly website by using the promo code PNLP20 or the link below. 

Link: https://learning.oreilly.com/get-learning/?code=PNLP20  

Episode Introduction:  

This is the 10th episode of the podcast and a really special one. I've got the authors of the Practical Natural Language Processing book. The book is a comprehensive guide to building, iterating and scaling real world NLP Systems. It is for anyone who is involved in any way in building NLP systems in industry - from software engineers to data scientists to ML engineers to product managers and business leaders. The book is already topping the charts on Amazon and has been endorsed by various experts from academia and industry.  

Episode Overview:  

So for this episode, I talk to the authors of the book - Sowmya, Bodhi, Anuj and Harshit.  We talk about the key ideas behind the book - about how it bridges the gap between theory and building practical ML/NLP solutions. We talk about the inspiration behind writing the book, how it stands out, how it has been structured, who can benefit from it and lots more. We also talk about the elephant in the room, GPT-3 and try to make sense of the hype around it and understand it's broader impact and how it positions us, as a community to leverage these systems on a wider scale.  

We also talk about the state of ML and NLP in general, about the many misconceptions and misinformed expectations that surround these fields in the context of the business of AI, and about how they've tried to incorporate this message in the book.  

Practical Natural Language Processing Book:  

Website: http://www.practicalnlp.ai/

Authors / Guests:  

Sowmya Vajjala: https://twitter.com/adyantalamadhya

Bodhisattwa Majumder: https://twitter.com/mbodhisattwa

Anuj Gupta: https://twitter.com/anujgupta82

Harshit Surana: https://twitter.com/surana_h

Connect with me 🙎🏻‍♂️:   

Website: https://mukulkhanna.github.io

Deep Neural Notebooks podcast 🎙: 

Youtube: www.youtube.com/channel/UC66w1T4oMv66Jn1LR5CW2yg

Anchor: www.anchor.fm/deep-neural-notebooks

Spotify: www.open.spotify.com/show/2eq1jD7V5K19aZUUJnIz5z

Apple Podcasts: www.podcasts.apple.com/in/podcast/deep-neural-notebooks/id1488705711

In this episode, I talk with Shalini De Mello, who is a Principal Research Scientist and Research Lead at NVIDIA. Her research interests are in computer vision and machine learning for human-computer interaction and smart interfaces.

At NVIDIA, she has developed technologies for gaze estimation, 2D and 3D head pose estimation, hand gesture recognition, face detection, video stabilization and GPU-optimized libraries for mobile computer vision. Her research has been focused on human-computer interaction in cars and has led to the development of NVIDIA’s innovative DriveIX product for smart AI-based automotive interfaces for future generations of cars.

Shalini received her Masters and PhD in Electrical and Computer Engineering from the University of Texas at Austin. She received a Bachelor of Engineering degree in Electronics and Electrical Communication Engineering from Punjab Engineering College.

In this episode, we talk about her journey - about how she got started with Computer Vision and Machine Learning, from her Bahelor's to her Master's in Biomedical Imaging to her PhD work on Human Face Recognition - about how her research interests shaped over the years. We also talk about Machine Learning inside the car, about her vision of using Machine Learning & Deep Learning for building smart assistive interfaces for inside the car, and about how that manifested into the DriveIX product that NVIDIA recently launched. Among other things, we talk about the importance of open-sourcing technology, about the future of autonomous and semi-autonomous vehicles, about the joys of learning something new everyday, about how to keep track of the every growing amount of research and much more. It was an absolute pleasure to talk with Shalini and learn from her research insights. I hope you like the conversation.

Shalini De Mello:

Twitter: https://twitter.com/shalinidemello

Website: https://research.nvidia.com/person/shalini-gupta

Links:

NVIDIA Drive IX: https://www.nvidia.com/en-us/self-driving-cars/drive-ix/, https://developer.nvidia.com/drive/drive-ix

Self-Supervised Viewpoint Learning From Image Collections: https://research.nvidia.com/publication/2020-03_Self-Supervised-Viewpoint-Learning

Multi-sensor System for Driver’s Hand-Gesture Recognition: https://research.nvidia.com/publication/hand-gesture-recognition-3d-convolutional-neural-networks

AI Co-Pilot: RNNs for Dynamic Facial Analysis: https://developer.nvidia.com/blog/ai-co-pilot-rnn-dynamic-facial-analysis/

Podcast links:

Spotify: https://tinyurl.com/yb6sn2rv

Apple Podcasts: https://tinyurl.com/y9hu7lzq

Google Podcasts: https://tinyurl.com/ybb8gxd5

Anchor.fm: https://tinyurl.com/ya98vk7b

Youtube: https://youtu.be/Hfz965mLuvM

Connect with me 🙎🏻‍♂️: 

Twitter: twitter.com/mkulkhanna

Instagram: instagram.com/mkulkhanna/

Deep Neural Notebooks podcast 🎙:

Youtube: www.youtube.com/channel/UC66w1T4oMv66Jn1LR5CW2yg

Anchor: www.anchor.fm/deep-neural-notebooks

Spotify: www.open.spotify.com/show/2eq1jD7V5K19aZUUJnIz5z

Apple Podcasts: www.podcasts.apple.com/in/podcast/deep-neural-notebooks/id1488705711

In this episode, I interview Varun Jampani, who is a Research Scientist at Google Research. You might recognise him from the renowned Super SloMo paper. His work lies at the intersection of Computer Vision and Machine Learning. His main focus is to leverage machine learning techniques for better inference in computer vision models.

Prior to joining Google, he was a research scientist at NVIDIA. He completed his PhD at the Max-Planck Institute (MPI) for Intelligent Systems. He is also a IIIT Hyderabad alum, where he did his Bachelor's and Master's.  

In this episode, we talk about his journey — from his Bachelor's and Masters at IIIT Hyderabad to his PhD at MPI, about how his research has shaped over the years, about his focus on always asking good research questions and tackling fundamental problems in Computer Vision as a whole.  We also talk about the SuperSloMo paper, about how it started, the key design decisions that were taken and the challenges faced in the process. If there's one thing that you are likely to take away from this conversation, it is the importance of asking good research questions and letting that drive your learning and research.  

Guest:  

Varun Jampani: https://varunjampani.github.io/  

Links:  

CVPR 2020 Novel View Synthesis Tutorial: https://www.youtube.com/watch?v=OEUHalxanuc

Episode links:

Spotify: https://tinyurl.com/y7u98d7m

Connect with me 🙎🏻‍♂️:   

Twitter: twitter.com/mkulkhanna

Deep Neural Notebooks podcast 🎙:

Youtube: www.youtube.com/channel/UC66w1T4oMv66Jn1LR5CW2yg

In the seventh episode of Deep Neural Notebooks, I interview Shimon Whiteson.  

Shimon sir is a Computer Science Professor at the University of Oxford, where he leads the Whiteson Research Lab. He is also a Data Scientist at Waymo (formerly the Google Self Driving Car Project). His research specialises in Reinforcement Learning (RL), Cooperative Multi-Agent RL, to be precise.   

So this interview is all in the context of Reinforcement Learning. We talk about his journey  - how he started with Machine Learning & RL. I ask him about his thoughts on the state of RL - about how the field has progressed and changed since he started, about how it has become so popular in the last few years, and about the challenges being faced.  

We also talk about his research at Waymo, about recent projects from his lab, and about the scope and future of telepresence robots, one of which was developed under his guidance. We also talk about the infamous Reward Hypothesis in the context of RL and Philosophy. In the end, he also shares some advice for people starting out with RL.  

Links:  

- Shimon Whiteson: https://twitter.com/shimon8282 

- Whiteson Research Lab (WhiRL): http://whirl.cs.ox.ac.uk/ 

- Teresa Robot: https://whirl.cs.ox.ac.uk/teresa/ 

- RL workshop at Machine Learning Summer School, Moscow: https://www.youtube.com/watch?v=RAw0Chs7QKA 

- The Reward Hypothesis: http://incompleteideas.net/rlai.cs.ualberta.ca/RLAI/rewardhypothesis.html

Timestamps:

03:42 Beginnings in Computer Science06:13 Beginnings in ML

07:15 PhD at UT Austin

10:40 Intersection of Neuroevolution and RL

14:10 Research directions since PhD

16:35 State of RL

20:33 Simulation for RL

22:07 Research at Waymo

25:30 Multi-agent RL

33:25 Recent projects at WhiRL

41:30 Teresa project and Telepresence Robots

48:08 Bottlenecks for RL and Robotics

49:45 End-goal for RL, Human-level Intelligence

53:45 What do you find most fascinating about your research?

55:38 RL & Philosophy

1:01:20 Keeping up with latest research

1:03:28 Advice for beginners

Podcast links :

Youtube: https://youtu.be/bbrYZDgPI9M

Apple Podcasts:  https://apple.co/2TLUZ0y

Google Podcasts: https://bit.ly/2TIyvh6

Spotify:  https://open.spotify.com/episode/3936aEvSwsIhfwQfURmDb9

Anchor: https://bit.ly/3gpMi65

Connect:  

Twitter: https://twitter.com/mkulkhanna 

Website: https://mukulkhanna.co 

LinkedIn: https://linkedin.com/in/mukulkhanna/

In the fifth episode of the Deep Neural Notebooks podcast, I interview Professor, Author, Sitarist and Composer, Srinivas Reddy.  

He is a Guest Professor of South and Southeast Asian Studies at IIT Gandhinagar and Visiting Assistant Professor of Religious Studies and Contemplative Studies at Brown University. He lives in Rhode Island and spends his time performing, teaching and conducting research around the world.  He is an author of 4 books, 3 of which are translations of Telugu and Sanskrit texts; and the latest book that came out this March is a critical biography of Krishnadevaraya of Vijayanagara, called RAYA.  

This episode is more in the realms of spirituality, in the context of South Asian philosophy and the Bhagwad Gita. We discuss some of his favourite verses and chapters from the Gita, about the deep underlying wisdom, and how we can make sense of them in today's day and age.  I ask him about his thoughts on religion, consciousness, meditation, his passion for music and much more.  He also talks about how India never really recovered from colonialisation and how the quest for adoption of Western ideologies has left us devoid of any appreciation for our cultural heritage, and how unconditioning ourselves is the first step in moving forward on our spiritual path.   

Keep in mind that this episode was recorded in the first week of January, wayyy before the pandemic outbreak; so some of his advice like going out for a walk on a beautiful sunny day was shared in a pre-COVID-19 context.  I hope you enjoy the episode, and learn something valuable from it. If you do, please leave a thumb's up or a 5-star rating depending on the streaming platform.  

Links: 

Srinivas Reddy: https://www.srinivasreddy.org/ 

Books: 

- Giver of the Worn Garland (Amuktamalyada) - http://www.tinyurl.com/amukta

Concerts:

Albums:

Podcast links -  

- Youtube: https://youtu.be/JK2X_ZM4vHM

Connect with me:  

Twitter: https://twitter.com/mkulkhanna

In this episode, I interview Leslee Lazar, a cognitive neuroscientist and visual artist. He is a professor at IIT Gandhinagar, at the Centre for Cognitive and Brain Science, working on processing of tactile perception in the somatosensory cortex of the brain. He is passionate about art and design, and uses illustrations, graphic design, infographics, collages and photography to convey complex stories. Neuroscientist by day, visual artist by evening, his research interests include understanding creativity and perception of art from a Neuroscience point of view. He has some amazing artworks, illustrations and posters that I'd recommend you to check out on his Instagram and Tumblr accounts, links to which can be found below.  

In this episode, we talk about his journey - from Zoology to Neuroscience, his work on touch perception, about creativity, and about how we as humans share an innate appreciation for art and beauty. On the intersection of Computer Science and Neuroscience, I asked him about brain computer interfaces, like the ones being developed at Neuralink, and his thoughts on possibility of being able to model a digital brain one day.   In the end, he shares some advice for people taking interest in Neuroscience and a list of books people can refer to, to get started.  

This video was recorded before the lockdown in India.  

If you enjoyed the conversation or learned something valuable from it, please give it a thumb's up or a 5-star rating depending on the streaming platform - it really helps the channel and allows more people to discover it.  

List of Top Neuroscience books for people starting out: 

Links:  

Leslee Lazar -  

Website: https://lesleelazar.com 

Twitter: https://twitter.com/leslee_lazar 

Instagram: https://instagram.com/dull_eye_llama 

Tumblr: https://lesleelazar.tumblr.com  

Centre for Cognitive and Brain Science: https://cogs.iitgn.ac.in  

Nature article on COVID-19: https://nature.com/articles/s41591-020-0820-9?fbclid=IwAR2xZouI4JVsZdFbIqLixXC-XC8mxMCQ_Inw2znKGcgrdeUW6kVzZQB_VVs  

Podcast links -  

Youtube: https://youtu.be/X2V5z6J6J1o 

Apple Podcasts: https://apple.co/2V22kZO

Google Podcasts: https://bit.ly/3aItP1o

Spotify: https://open.spotify.com/episode/4TuU8kZBCM2KWevHBAbpNj 

Anchor: https://anchor.fm/deep-neural-notebooks/episodes/DNN-5-Neuroscience--Art-and-Creativity--Leslee-Lazar-ecaqoc

Connect:  

Twitter: https://twitter.com/mkulkhanna 

Website: https://mukulkhanna.co 

LinkedIn: https://linkedin.com/in/mukulkhanna/

In the fourth episode of Deep Neural Notebooks, I talk with Mr. Mukul Pandya. 

He is the the editor-in-chief and executive director of the Knowledge@Wharton, the online business analysis journal of the Wharton School of the University of Pennsylvania. He is a Senior Fellow in the Wharton School’s Management Department. He is a winner of four awards for investigative journalism and author of three books, Pandya has more than 30 years of experience as a writer and editor. His articles have appeared in the Wall Street Journal, the New York Times, The Economist and other publications. He holds a master’s degree in economics from the University of Bombay. 

We talk about many things - about his journey, his beginnings, the story of Knowledge@Wharton, about his experiences of interviewing legends like Adam Grant, the founder of Intel and Dr. APJ Abdul Kalam, about the importance and drawbacks of technology in the conquest for knowledge, about figuring out the right balance and utilising technology to exceed our abilities, about the core values that would be of utmost importance in the coming decades of unprecedented transformations caused by automation and AI, and about what he feels is the most important aspect of being a successful leader.

In the end, Mukul sir also shares some beautiful advice for individuals that seek to learn and share knowledge. Mukul sir is an absolute gentleman, probably the most humble, down-to-earth person I have ever come across. 

I hope you like our conversation. 

Links-

Mukul Pandya - https://twitter.com/MPandya

Knowledge@Wharton - https://knowledge.wharton.upenn.edu/ 

Knowledge@Wharton Podcast - https://knowledge.wharton.upenn.edu/category/podcasts/ 

Interview with Dr. APJ Abdul Kalam - https://www.youtube.com/watch?v=laGZaS4sdeU  

Podcast links -   

Youtube: https://youtu.be/xK9QGuIBjSE 

Apple Podcasts:  https://apple.co/3blfwj6

Google Podcasts: https://bit.ly/2Ua67oG

Spotify: https://open.spotify.com/episode/3POqfBrBi19WtX66TgfGqK

Anchor:  https://anchor.fm/deep-neural-notebooks/episodes/DNN-4-Knowledge--Tech--Leadership--KnowledgeWharton--Interview-with-Mukul-Pandya-ebqs4k/a-a1oj46o

Connect:  

Website: https://mukulkhanna.co 

Twitter: https://twitter.com/mkulkhanna

LinkedIn: https://linkedin.com/in/mukulkhanna/

In the third episode of Deep Neural Notebooks, I talk with Michael Droettboom, from Mozilla.

Michael is a Staff Data Engineer at Mozilla, where he works on managing the telemetry from Mozilla products, that can be used for improving the user experience and the product itself, while respecting data privacy and making sure that instead of snarfing it all up, only the absolutely minimal data is gathered from the user. 

He specialises in imaging and data: sheet music, scientific visualization, astronomy, biomedical data and software telemetry.

He started his Open Source journey around 2007, contributing to Matplotlib, while he was a Senior Computer Scientist at the Space Telescope Science Institute. He later went on to become a lead developer of Matplotlib, carrying on the vision of Matplotlib’s original author, John Hunter.

In this episode, we talk about his beginnings - his education in Computers’ and Music, his experience in Astronomy, working at the Space Telescope Science Institute and how he got into contributing to Open Source for Matplotlib. We also talk about his role at Mozilla, the importance of data privacy and the amazing project that he is currently working on - Pyodide, a tool to empower Data Science in the Browser.

Michael also shares some advice for beginners trying to get into Data Science and communities just starting out with Open Source software.

Michael Droettboom: https://twitter.com/MDroettboom , http://droettboom.com/

GLEAN: https://github.com/mozilla/glean

Pyodide: https://github.com/iodide-project/pyodide

Deep Neural Notebooks is a podcast where I like to discuss a multitude of topics, ranging from Deep Learning and Computer Vision to Neuroscience and Open Source Software, through conversations with experts about their thoughts on the state of their specialisations, how things fit into the bigger picture, their journey so far and the road ahead.

I believe that it is through conversations like these that we can boil down the essence of vast resources of knowledge and expertise into more consumable bits that can enrich our understanding of concepts and technologies that are shaping our world.

If you like the content, please subscribe to the channel and leave a thumbs up, or a 5-star rating, depending on the streaming platform.

Youtube: https://www.youtube.com/channel/UC66w1T4oMv66Jn1LR5CW2yg

Apple Podcasts: https://podcasts.apple.com/us/podcast/deep-neural-notebooks/id1488705711?uo=4

Spotify: https://open.spotify.com/show/2eq1jD7V5K19aZUUJnIz5z

Google Podcasts: https://www.google.com/podcasts?feed=aHR0cHM6Ly9hbmNob3IuZm0vcy8xMDZkYzIzOC9wb2RjYXN0L3Jzcw==

Anchor: https://anchor.fm/deep-neural-notebooks

Connect:

Website: https://mukulkhanna.github.io

LinkedIn: https://linkedin.com/in/mukulkhanna/

Twitter: https://twitter.com/mkulkhanna

In the second episode of Deep Neural Notebooks, I talk with Thomas Wolf. 

Thomas is the Chief Science Officer at HuggingFace, a Brooklyn based start-up that aims at building the first truly social artificial intelligence. His interesting background includes a PhD in Quantum Physics and a degree in Law, followed by a five-year long career as a Patent Attorney, before his passion for Science brought him to HuggingFace. Thomas, and his team at HuggingFace believes in the power of Open Source and have been active contributors on Github, sharing their research and progress, allowing the development of technology for the better. 

In this episode, we talk about his diverse career background- his journey from Physics to Law to Deep Learning and Conversational Agents. We also talk about the vision at HuggingFace, the challenges in building a long-term, companion-like Conversational Agent, the state of Natural Language Processing, and how we can do better. 

Thomas Wolf: https://twitter.com/Thom_Wolf , http://thomwolf.io/

HuggingFace: https://huggingface.co

HuggingFace’s Open Source Chatbot Repository (NeurIPS’18 ConvAI2 submission): https://github.com/huggingface/transfer-learning-conv-ai

Magic-Sand: https://github.com/thomwolf/Magic-Sand

Magic-Sand tutorial: https://imgur.com/gallery/Q86wR

Deep Neural Notebooks:

Deep Neural Notebooks is a podcast where I like to discuss a multitude of topics, ranging from Deep Learning and Computer Vision to Neuroscience and Open Source Software, through conversations with experts about their thoughts on the state of their specialisations, how things fit into the bigger picture, their journey so far and the road ahead.

I believe that it is through conversations like these that we can boil down the essence of vast resources of knowledge and expertise into more consumable bits that can enrich our understanding of concepts and technologies that are shaping our world.

If you like the content, please subscribe to the channel and leave a thumbs up, or a 5-star rating, depending on the streaming platform.

Youtube: https://www.youtube.com/channel/UC66w1T4oMv66Jn1LR5CW2yg

Apple Podcasts: https://podcasts.apple.com/us/podcast/deep-neural-notebooks/id1488705711?uo=4

Spotify: https://open.spotify.com/show/2eq1jD7V5K19aZUUJnIz5z

Google Podcasts: https://www.google.com/podcasts?feed=aHR0cHM6Ly9hbmNob3IuZm0vcy8xMDZkYzIzOC9wb2RjYXN0L3Jzcw==

Anchor: https://anchor.fm/deep-neural-notebooks

Connect:

Website: https://mukulkhanna.github.io

LinkedIn: https://linkedin.com/in/mukulkhanna/

Twitter: https://twitter.com/mkulkhanna

In the first episode of the Deep Neural Notebooks podcast, I talk with Shanmuganathan Raman, who is an Associate Professor of Electrical  & Computer Science Engineering at IIT Gandhinagar. He has been my mentor for my research internship at IIT. He has obtained his MTech and PhD degrees from IIT Bombay and was a post-doc research associate at Department of Electrical Engineering, Indian Institute of Science (IISc), Bangalore. His PhD thesis on 'Low Dynamic Range solution to High Dynamic Range Imaging problem' received the IIT Bombay Excellence in PhD Thesis Work' Award. His research interests include Computer Vision, Computational Photography, Machine Learning and Computer Graphics.

In this episode we discuss his educational background, his theses, the state of HDR imaging and computational photography. We also talk about deep learning, the scope of traditional algorithms in a DL-minded society and the road ahead.

Relevant links:

Prof. Shanmuganathan Raman: people.iitgn.ac.in/~shanmuga/

Deep Neural Notebooks:

Deep Neural Notebooks is a podcast where I like to discuss a multitude of topics, ranging from Deep Learning and Computer Vision to Neuroscience and Open Source Software, through conversations with experts about their thoughts on the state of their specialisations, how things fit into the bigger picture, their journey so far and the road ahead.

I believe that it is through conversations like these that we can boil down the essence of vast resources of knowledge and expertise into more consumable bits that can enrich our understanding of concepts and technologies that are shaping our world.

If you like the content, please subscribe to the channel and leave a thumbs up, or a star rating, depending on the streaming platform.

Apple Podcasts: podcasts.apple.com/in/podcast/deep-neural-notebooks/id1488705711

Spotify: open.spotify.com/show/2eq1jD7V5K19aZUUJnIz5z

Anchor: anchor.fm/deep-neural-notebooks

Connect:

LinkedIn: linkedin.com/in/mukulkhanna/

Twitter: twitter.com/mkulkhanna