Novel AI Approaches
• 1 min read 1 min
What I worked on
Explored non-DNN methods in AI and which ones appear at major conferences. Looked at publication trends and new neuron types used in research.
What I noticed
- Evolutionary computation has steady research output
- Novel neuron types continue to appear but rarely displace DNNs
- One-shot inference remains uncommon outside neural methods
”Aha” Moment
na
What still feels messy
Which alternative approaches have genuine long-term promise.
Next step
Read recent NeurIPS papers on evolutionary computation and energy-based models.