AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Recursive Binding on a Budget: Subspace Carving in Order-$p$ Tensor Memories

Travis Pence, Daisuke Yamada, Vikas Singh

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Differentiable optimization layers · 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
replace VSA binding with OSC subspace carving at 1000 bundlesmemory architecture · Synthetic memory, 1000 bundles, p=2retrieval accuracy (%): 13.18 -> 99.77Improved
use OSC with p=2 instead of best VSA d=4096memory architecture · Synthetic memory, 1000 bundlestotal memory (parameters): 4100096 -> 440000Improved
use OSC p=3 d=80 instead of HLB d=343000 at 10000 bundlesmemory architecture · Synthetic memory, 10000 bundles, p=3total memory (parameters): 3430343000 -> 2912000Improved
use OSC instead of TPR at small superposition N=50memory architecture · Min filler dimension for 99% accuracytotal parameters: 4K -> 4KNo clear effect
use OSC instead of TPR at large superposition N=10000memory architecture · Min filler dimension for 99% accuracytotal parameters: 6.9M -> 2.7MImproved
hold memory size constant and compare capacity with HLBOSC · Synthetic memory, p=3memory capacityHurt
use OSC instead of HLB for retrieval at small superpositionOSC · Retrieval time, N=100, B1speed (ms): 0.068 -> 0.149Hurt
use OSC instead of HLB for retrieval at large superpositionOSC · Retrieval time, N=10000, B2speed (ms): 1.189 -> 0.306Improved
increase codebook size at fixed superposition (ablation)OSC · Min filler dimension for 99% accuracyrequired filler dimensionNo clear effect
use OSC in extreme multi-label learning instead of VTB/MAP/HLBOSC · XML EURLEX-4KnDCG@5: 52.54 -> 59.70Improved
use OSC in extreme multi-label learning instead of VTBOSC · XML EURLEX-4.3KnDCG@5: 75.91 -> 71.59Hurt
use OSC in extreme multi-label learning instead of HLBOSC · XML DELICIOUS-200KnDCG@5: 38.80 -> 37.28Hurt

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