43 datasets found

The proposed Siamese-Predictor using EFM to fuse prior knowledge - Estimation Code. The proposed Estimation Code and accuracy of architectures are highly correlated based on our...

Structural re-parameterization technology provides a new idea to improve the performance of traditional convolutional networks.

The proposed method uses NASBench-201, DARTS, and MobileNet Space for experiments.

The dataset used in the paper to test the proposed Decomposable Winograd Method (DWM) for convolution acceleration.

Face anti-spoofing (FAS) task, neural architecture search (NAS), convolution, pooling, static-dynamic representation

Face anti-spoofing (FAS) task, neural architecture search (NAS), convolution, pooling, static-dynamic representation

MnasNet: In platform-aware neural architecture search for mobile

UNAS: Differentiable Architecture Search Meets Reinforcement Learning

The dataset used in the paper is NAS-Bench-101, a benchmark for neural architecture search.

The dataset used in the paper is ENAS Macro, a search space for neural architecture search.

The dataset used in the paper is RandWire, a randomly wired neural network search space.

The authors propose a proxyless NAS algorithm that can be used for efficient deployment.

The authors propose a multi-objective neural architecture search using reinforcement learning.

The authors propose a novel hardware-aware differentiable NAS algorithm for profiling the Pareto front in multi-objective problems.

Beta-DARTS: Beta-Decay Regularization for Differentiable Architecture Search

NAS-Bench-X11: A Benchmark for Neural Architecture Search

Bananas: Bayesian Optimization with Neural Architectures for Neural Architecture Search

Multi-conditioned Graph Diffusion for Neural Architecture Search

This paper presents a novel framework for data-free neural architecture search (NAS) without access to the original data.

Progressive Neural Architecture Search