Is it enough to report performance on a single dataset? Can we trust the reported improvement of new model architectures? We created 144 experiments for…
and
1
The current limitations of Graph Neural Networks. We continue our two part series on ML on Graphs, by asking: could graphs replace other domain specific…
and
3
Why Machine Learning on Graphs could be the future of AI. Many researchers are doubling down on scaling models like GPT-3 on an ever bigger corpus of…
and
3
Deep Reinforcement Learning is about learning from mistakes - although all machine learning algorithms depend on making sense of the magnitude of error…
and
2
This is Applied Exploration, a newsletter about SOTA Machine Learning in practice.
and
2