Fine-tuning from pre-trained ImageNet models has become the de-facto standard for various computer vision tasks. Current practices for fine-tuning typically involve selecting an ad-hoc choice of hyperparameters and keeping them fixed to values normally used for training from scratch. This paper re-examines several common practices of setting hyperparameters for fine-tuning. Our findings are based on extensive empirical evaluation for fine-tuning on various transfer learning benchmarks. (1) While prior works have thoroughly investigated learning rate and batch size, momentum for fine-tuning is a relatively unexplored parameter. We find that the value of momentum also affects fine-tuning performance and connect it with previous theoretical findings. (2) Optimal hyperparameters for fine-tuning, in particular, the effective learning rate, are not only dataset dependent but also sensitive to the similarity between the source domain and target domain. This is in contrast to hyperparameters for training from scratch. (3) Reference-based regularization that keeps models close to the initial model does not necessarily apply for "dissimilar" datasets. Our findings challenge common practices of fine-tuning and encourages deep learning practitioners to rethink the hyperparameters for fine-tuning.
Federal Technology University（联邦技术大学）；University of São Paulo（圣保罗大学）；Universidade Estadual Paulista（保利斯塔州立大学）；Eindhoven University of Technology（埃因霍温理工大学）
For many machine learning algorithms, predictive performance is critically affected by the hyperparameter values used to train them. However, tuning these hyperparameters can come at a high computational cost, especially on larger datasets, while the tuned settings do not always significantly outperform the default values. This paper proposes a recommender system based on meta-learning to identify exactly when it is better to use default values and when to tune hyperparameters for each new dataset. Besides, an in-depth analysis is performed to understand what they take into account for their d…
Bayesian optimization (BO) is a sample efficient approach to automatically tune the hyperparameters of machine learning models. In practice, one frequently has to solve similar hyperparameter tuning problems sequentially. For example, one might have to tune a type of neural network learned across a series of different classification problems. Recent work on multi-task BO exploits knowledge gained from previous tuning tasks to speed up a new tuning task. However, previous approaches do not account for the fact that BO is a sequential decision making procedure. Hence, there is in general a misma…
Despite the evolution of Convolutional Neural Networks (CNNs), their performance is surprisingly dependent on the choice of hyperparameters. However, it remains challenging to efficiently explore large hyperparameter search space due to the long training times of modern CNNs. Multi-fidelity optimization enables the exploration of more hyperparameter configurations given budget by early termination of unpromising configurations. However, it often results in selecting a sub-optimal configuration as training with the high-performing configuration typically converges slowly in an early phase. In t…
Peking University（北京大学）；Tencent Inc.（腾讯公司）；ETH Zürich（苏黎世联邦理工学院）；Beijing Institute of Technology（北京理工大学）；Peking University (Qingdao)（北京大学（青岛））
The tuning of hyperparameters becomes increasingly important as machine learning (ML) models have been extensively applied in data mining applications. Among various approaches, Bayesian optimization (BO) is a successful methodology to tune hyper-parameters automatically. While traditional methods optimize each tuning task in isolation, there has been recent interest in speeding up BO by transferring knowledge across previous tasks. In this work, we introduce an automatic method to design the BO search space with the aid of tuning history from past tasks. This simple yet effective approach can…
Alibaba Group（阿里巴巴集团）；National University of Singapore（新加坡国立大学）
Parameter-efficient transfer learning (PETL) based on large-scale pre-trained foundation models has achieved great success in various downstream applications. Existing tuning methods, such as prompt, prefix, and adapter, perform task-specific lightweight adjustments to different parts of the original architecture. However, they take effect on only some parts of the pre-trained models, i.e., only the feed-forward layers or the self-attention layers, which leaves the remaining frozen structures unable to adapt to the data distributions of downstream tasks. Further, the existing structures are st…
Alleviating Hyperparameter-Tuning Burden in SVM Classifiers for Pulmonary Nodules Diagnosis with Multi-Task Bayesian Optimization
Yau Mathematical Sciences Center Tsinghua University（姚数学科学中心清华大学）；The First Affiliated Hospital of Guangzhou Medical University（广州医科大学第一附属医院）；China State Key Laboratory of Respiratory Disease（中国呼吸疾病国家重点实验室）；Academy of Mathematics and Systems Science Chinese Academy of Sciences（数学科学学院中国科学院）；University of Chinese Academy of Sciences（中国科学院大学）；Department of Thoracic Surgery and Oncology The First Affiliated Hospital of Guangzhou Medical University（胸外科与肿瘤科广州医科大学第一附属医院）
The ImageNet pre-training initialization is the de-facto standard for object detection. He et al. found it is possible to train detector from scratch(random initialization) while needing a longer training schedule with proper normalization technique. In this paper, we explore to directly pre-training on target dataset for object detection. Under this situation, we discover that the widely adopted large resizing strategy e.g. resize image to (1333, 800) is important for fine-tuning but it's not necessary for pre-training. Specifically, we propose a new training pipeline for object detection tha…
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
Despite achieving state-of-the-art zero-shot performance, existing vision-language models still fall short of few-shot transfer ability on domain-specific problems. Classical fine-tuning often fails to prevent highly expressive models from exploiting spurious correlations. Although model-agnostic meta-learning (MAML) presents as a natural alternative for few-shot transfer learning, the expensive computation due to implicit second-order optimization limits its use on large-scale vision-language models such as CLIP. While much literature has been devoted to exploring alternative optimization str…
