AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Differentiable Weightless Controllers: Learning Logic Circuits for Continuous Control

Fabian Kresse, Christoph H. Lampert

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
use DWC instead of full-precision networkSAC policy · Ant-v4return: 5598.251 -> 5677.335Improved
use DWC instead of full-precision networkSAC policy · HalfCheetah-v4return: 11528.543 -> 7548.604Hurt
use DWC instead of full-precision networkSAC policy · Hopper-v4return: 2796.813 -> 3119.865Improved
use DWC instead of quantized networkSAC policy · Walker2d-v4return: 4656.477 -> 5024.956Improved
use DWC instead of quantized network on FPGASAC policy · Ant FPGA energy per actionenergy per action (J): 8.93e-7 -> 2.25e-09Improved
increase LUT layer width from 256 to 1024 (ablation)DWC · HalfCheetah-v4return: 7054.791 -> 7548.604Improved
increase LUT layer width from 256 to 1024 (ablation)DWC · Ant-v4return: 4542.495 -> 5677.335Improved
increase layer width to 16,384 LUTs and 255 input levels (ablation)DWC · HalfCheetah-v4return: 7.5k -> above 10.3kImproved
reduce LUT inputs k from 6 to 2 (ablation)DWC · Ant, Hopper, Humanoid, Walker2dreturnNo clear effect
use more thermometer input bits (5 to 63) (ablation)DWC · MuJoCo tasksreturnImproved
add more LUT layers beyond one (ablation)DWC (D=1024) · MuJoCo tasksreturnNo clear effect
use DWC with DDPG instead of FPDDPG policy · HalfCheetah-v4return: 11278.164 -> 7293.430Hurt

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