The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) AutoML for Natural Language Processing with Abhishek Thakur - #475
Apr 15, 2021
Abhishek Thakur, a machine learning engineer at Hugging Face and the world’s first quadruple Kaggle Grandmaster, shares insights from his fascinating journey. He discusses his evolution in Kaggle competitions, emphasizing practical skills gained along the way. Abhishek dives into his work on AutoNLP, revealing its goals and how it stacks up against handcrafted models. He also highlights key lessons in NLP techniques and the importance of blending theory with practice, alongside his experiences writing his book, Approaching (Almost) Any Machine Learning Problem.
AI Snips
Chapters
Books
Transcript
Episode notes
Abhishek's Kaggle Beginning
- Abhishek Thakur's machine learning journey started with an internship using Random Forest.
- He transitioned from MATLAB to Python after his first Kaggle competition on emotion recognition.
Kaggle's Real-World Value
- Kaggle competitions offer clean data and a learning environment.
- Apply Kaggle knowledge to real-world scenarios and learn by competing and collaborating.
Key Kaggle Learnings
- Handling categorical variables effectively is crucial.
- Competitions teach valuable time management and diverse problem-solving skills.
