Heinrich van Deventer

Physicist that landed in AI and Machine Learning. Computer Science Student. Adapting ideas from functional analysis and physics to machine learning.

[Anna Sergeevna Bosman @annaearwen]

It's happening again - we're organising the next #Indabax2024, and it'll be the best and the biggest one yet! Join us - and bring your posters & talks🤩

[Anna Sergeevna Bosman @annaearwen]

New preprint by @hpdeventer: https://t.co/JsnV9gm6Na
Why do NNs struggle to learn tasks in sequence? Because training a model on a domain subset yields non-local changes on other domain subsets. What if we construct a model where the changes are local? Read the paper to find out!

[Anna Sergeevna Bosman @annaearwen]

Congratulations to @hpdeventer for being selected for this prestigious fellowship ✨I am excited to be involved as a supervisor @UPTuks. Looking forward to all the discoveries, Heinrich!

#CIRG #MachineLearning

[Heinrich van Deventer @hpdeventer]

@sarahookr Yes.

[Heinrich van Deventer @hpdeventer]

Excited to attend @AIHackTunisia as a participant!

[Heinrich van Deventer @hpdeventer]

@thegautamkamath But all contributions are worthwhile in theory.

[Gautam Kamath @thegautamkamath]

Theory researchers be like

[Heinrich van Deventer @hpdeventer]

@sarahookr @DeepIndaba @orevaahia @vykthur @dansitu The ML efficiency workshop was fantastic! It was a lot of fun and thanks for all the effort in making it a reality.

[Bonaventure F. P. Dossou @bonadossou]

@MasakhaneNLP got at least 1 winner in awards-winning @DeepIndaba 🤩

[Heinrich van Deventer @hpdeventer]

I tackled catastrophic forgetting, and instead uncovered an efficient universal function approximator. Check out the current draft of my master's work.

Thanks to the organizers and sponsors for making the #DLIndaba2022 such a wonderful experience!
https://t.co/xN6zXfSHNJ