# TF-NAS: Rethinking Three Search Freedoms of Latency-Constrained Differentiable Neural Architecture Search

arXiv: https://arxiv.org/abs/2008.05314 · HTML: https://mlsanity.com/paper/2008.05314/

## Key facts

- Cited by 21 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Aug 2020.

## Built on by (most cited first, 21 of 21)

- [Investigating Bi-Level Optimization for Learning and Vision from a Unified Perspective: A Survey and Beyond](https://mlsanity.com/paper/2101.11517/) (2021)
- [Dive into Ambiguity: Latent Distribution Mining and Pairwise Uncertainty Estimation for Facial Expression Recognition](https://mlsanity.com/paper/2104.00232/) (2021)
- [A Comprehensive Survey on Hardware-Aware Neural Architecture Search](https://mlsanity.com/paper/2101.09336/) (2021)
- [FaceX-Zoo: A PyTorch Toolbox for Face Recognition](https://mlsanity.com/paper/2101.04407/) (2021)
- [HR-NAS: Searching Efficient High-Resolution Neural Architectures with Lightweight Transformers](https://mlsanity.com/paper/2106.06560/) (2021)
- [CM-NAS: Cross-Modality Neural Architecture Search for Visible-Infrared Person Re-Identification](https://mlsanity.com/paper/2101.08467/) (2021)
- [A General Descent Aggregation Framework for Gradient-based Bi-level Optimization](https://mlsanity.com/paper/2102.07976/) (2021)
- [Locally Free Weight Sharing for Network Width Search](https://mlsanity.com/paper/2102.05258/) (2021)
- [Face.evoLVe: A High-Performance Face Recognition Library](https://mlsanity.com/paper/2107.08621/) (2021)
- [HardCoRe-NAS: Hard Constrained diffeRentiable Neural Architecture Search](https://mlsanity.com/paper/2102.11646/) (2021)
- [Searching by Generating: Flexible and Efficient One-Shot NAS with Architecture Generator](https://mlsanity.com/paper/2103.07289/) (2021)
- [Auto-NBA: Efficient and Effective Search Over the Joint Space of Networks, Bitwidths, and Accelerators](https://mlsanity.com/paper/2106.06575/) (2021)
- [Reconsidering CO2 emissions from Computer Vision](https://mlsanity.com/paper/2104.08702/) (2021)
- [TabNAS: Rejection Sampling for Neural Architecture Search on Tabular Datasets](https://mlsanity.com/paper/2204.07615/) (2022)
- [Low-Mid Adversarial Perturbation against Unauthorized Face Recognition System](https://mlsanity.com/paper/2206.09410/) (2022)
- [FreeAugment: Data Augmentation Search Across All Degrees of Freedom](https://mlsanity.com/paper/2409.04820/) (2024)
- [BINAS: Bilinear Interpretable Neural Architecture Search](https://mlsanity.com/paper/2110.12399/) (2021)
- [AutoAL: Automated Active Learning with Differentiable Query Strategy Search](https://mlsanity.com/paper/2410.13853/) (2024)
- [Enabling Hard Constraints in Differentiable Neural Network and Accelerator Co-Exploration](https://mlsanity.com/paper/2301.09312/) (2023)
- [Pushing Joint Image Denoising and Classification to the Edge](https://mlsanity.com/paper/2409.08943/) (2024)
- [Exploring Deep-to-Shallow Transformable Neural Networks for Intelligent Embedded Systems](https://mlsanity.com/paper/2512.19731/) (2025)

## Abstract (arXiv)

With the flourish of differentiable neural architecture search (NAS), automatically searching latency-constrained architectures gives a new perspective to reduce human labor and expertise. However, the searched architectures are usually suboptimal in accuracy and may have large jitters around the target latency. In this paper, we rethink three freedoms of differentiable NAS, i.e. operation-level, depth-level and width-level, and propose a novel method, named Three-Freedom NAS (TF-NAS), to achieve both good classification accuracy and precise latency constraint. For the operation-level, we present a bi-sampling search algorithm to moderate the operation collapse. For the depth-level, we introduce a sink-connecting search space to ensure the mutual exclusion between skip and other candidate operations, as well as eliminate the architecture redundancy. For the width-level, we propose an elasticity-scaling strategy that achieves precise latency constraint in a progressively fine-grained manner. Experiments on ImageNet demonstrate the effectiveness of TF-NAS. Particularly, our searched TF-NAS-A obtains 76.9% top-1 accuracy, achieving state-of-the-art results with less latency. The total search time is only 1.8 days on 1 Titan RTX GPU. Code is available at https://github.com/AberHu/TF-NAS.