AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Neural Attentive Circuits

Martin Weiss, Nasim Rahaman, Francesco Locatello, Christopher Pal, Yoshua Bengio, Bernhard Schölkopf, Li Erran Li, Nicolas Ballas

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Dense features and foundation priors · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add ModFC layers to Perceiver IO (ablation)Perceiver IO · Tiny-ImageNet validationAcc@1: 58.89 -> 58.91No clear effect
Add ModFC layers to Perceiver IO (ablation)Perceiver IO · Tiny-ImageNet-RAcc@1: 17.57 -> 18.01No clear effect
Add a sparse graph sampled from a frozen prior (ablation)Perceiver IO + ModFC · Tiny-ImageNet validationAcc@1: 58.91 -> 58.31Hurt
Add a sparse graph sampled from a frozen prior (ablation)Perceiver IO + ModFC · Tiny-ImageNet-RAcc@1: 18.01 -> 18.50Improved
Learn the connectivity graph and module codes jointly (ablation)Perceiver IO + ModFC + sparse graph · Tiny-ImageNet validationAcc@1: 58.31 -> 60.76Improved
Learn the connectivity graph and module codes jointly (ablation)Perceiver IO + ModFC + sparse graph · Tiny-ImageNet-RAcc@1: 18.50 -> 19.52Improved
Use a scale-free rather than ring-of-cliques graph prior (ablation)NAC · Tiny-ImageNet validationAcc@1: 60.54 -> 60.76Improved
Use a ring-of-cliques rather than scale-free graph prior (ablation)NAC · Tiny-ImageNet-RAcc@1: 19.52 -> 20.03Improved
Use NAC rather than Perceiver IOPoint-cloud classification · Point-cloud classification testtest accuracy: 77.4% -> 83.0%Improved
Use NAC rather than Perceiver IOListOps · ListOps testtest accuracy: 39.70% -> 41.40%Improved
Use NAC rather than Perceiver IOASCII-byte text classification · Text classification from ASCII bytes testtest accuracy: 66.50% -> 68.18%Improved
Drop more than 80% of the least-connected modules at inference (ablation)NAC · Tiny-ImageNetinference speedup: ~8x speedup; less than 3% performance lossImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy