Author name cluster

Ran Ma

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

7 papers

2 author rows

ICML Conference 2025 Conference Paper

Adapter Naturally Serves as Decoupler for Cross-Domain Few-Shot Semantic Segmentation

Jintao Tong
Ran Ma
Yixiong Zou
Guangyao Chen
Yuhua Li 0003
Ruixuan Li 0001

Cross-domain few-shot segmentation (CD-FSS) is proposed to first pre-train the model on a source-domain dataset with sufficient samples, and then transfer the model to target-domain datasets where only a few training samples are available for efficient finetuning. There are majorly two challenges in this task: (1) the domain gap and (2) finetuning with scarce data. To solve these challenges, we revisit the adapter-based methods, and discover an intriguing insight not explored in previous works: the adapter not only helps the fine-tuning of downstream tasks but also naturally serves as a domain information decoupler. Then, we delve into this finding for an interpretation, and we find the model’s inherent structure could lead to a natural decoupling of domain information. Building upon this insight, we propose the Domain Feature Navigator (DFN), which is a structure-based decoupler instead of loss-based ones like current works, to capture domain-specific information, thereby directing the model’s attention towards domain-agnostic knowledge. Moreover, to prevent the potential excessive overfitting of DFN during the source-domain training, we further design the SAM-SVN method to constrain DFN from learning sample-specific knowledge. On target domains, we freeze the model and fine-tune the DFN to learn knowledge specific to target domains. Extensive experiments demonstrate that our method surpasses the state-of-the-art method in CD-FSS significantly by 2. 69% and 4. 68% average MIoU in 1-shot and 5-shot scenarios, respectively.

Details

AAAI Conference 2025 Conference Paper

Reconstruction Target Matters in Masked Image Modeling for Cross-Domain Few-Shot Learning

Ran Ma
Yixiong Zou
Yuhua Li
Ruixuan Li

Cross-Domain Few-Shot Learning (CDFSL) requires the model to transfer knowledge from the data-abundant source domain to data-scarce target domains for fast adaptation, where the large domain gap makes CDFSL a challenging problem. Masked Autoencoder (MAE) excels in effectively using unlabeled data and learning image’s global structures, enhancing model generalization and robustness. However, in the CDFSL task with significant domain shifts, we find MAE even shows lower performance than the baseline supervised models. In this paper, we first delve into this phenomenon for an interpretation. We find that MAE tends to focus on low-level domain information during reconstructing pixels while changing the reconstruction target to token features could mitigate this problem. However, not all features are beneficial, as we then find reconstructing high-level features can hardly improve the model’s transferability, indicating a trade-off between filtering domain information and preserving the image’s global structure. In all, the reconstruction target matters for the CDFSL task. Based on the above findings and interpretations, we further propose Domain-Agnostic Masked Image Modeling (DAMIM) for the CDFSL task. DAMIM includes an Aggregated Feature Reconstruction module to automatically aggregate features for reconstruction, with balanced learning of domain-agnostic information and images’ global structure, and a Lightweight Decoder module to further benefit the encoder’s generalizability. Experiments on four CDFSL datasets demonstrate that our method achieves state-of-the-art performance.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

Attention Temperature Matters in ViT-Based Cross-Domain Few-Shot Learning

Yixiong Zou
Ran Ma
Yuhua Li
Ruixuan Li

Cross-domain few-shot learning (CDFSL) is proposed to transfer knowledge from large-scale source-domain datasets to downstream target-domain datasets with only a few training samples. However, Vision Transformer (ViT), as a strong backbone network to achieve many top performances, is still under-explored in the CDFSL task in its transferability against large domain gaps. In this paper, we find an interesting phenomenon of ViT in the CDFSL task: by simply multiplying a temperature (even as small as 0) to the attention in ViT blocks, the target-domain performance consistently increases, even though the attention map is downgraded to a uniform map. In this paper, we delve into this phenomenon for an interpretation. Through experiments, we interpret this phenomenon as a remedy for the ineffective target-domain attention caused by the query-key attention mechanism under large domain gaps. Based on it, we further propose a simple but effective method for the CDFSL task to boost ViT's transferability by resisting the learning of query-key parameters and encouraging that of non-query-key ones. Experiments on four CDFSL datasets validate the rationale of our interpretation and method, showing we can consistently outperform state-of-the-art methods. Our codes are available at https: //github. com/Zoilsen/Attn Temp CDFSL.

PDF Details DOI

TCS Journal 2023 Journal Article

Online scheduling with deterioration and unexpected processor breakdown

Sainan Guo
Ran Ma
Yuefang Sun
Xiaoyan Zhang
Yong Zhang

In this paper, we investigate a single processor scheduling problem where the breakdown interval emerges on the processor in an online way. In particular, in our setting, the relevant data of each job J k, including the release date r k and the deteriorating rate b k, is known in advance. Furthermore, for each job J k, its processing time p k is a simple linear function of its starting time S k, i. e. , p k = b k S k. We are interested in scheduling the jobs to minimize the makespan. In addition, we indicate that, there exists no online algorithm such that the competitive ratio is less than 1 + b min, where b min = min 1 ≤ k ≤ n ⁡ b k. Moreover, we propose an online algorithm with a competitive ratio of 1 + b max, where b max = max 1 ≤ k ≤ n ⁡ b k.

Details DOI

TCS Journal 2022 Journal Article

An optimal online algorithm for single-processor scheduling problem with learning effect

Ran Ma
Sainan Guo
Xiaoyan Zhang

This paper deals with the canonical single-processor online scheduling problem with the position-based learning effect. Specially speaking, a round of jobs arriving online over time will be processed on a single processor. Noticeably, in this model, for each job J k, the actual processing time p k l is defined as a power function of its position l, i. e. , p k l = p k l β, where p k indicates its normal processing time and β ≤ 0 is the learning index. Our goal is to make the sum of completion times as small as possible. For this problem, we testify that there is no online algorithm with a competitive ratio of less than 2. Most notably, we design an online algorithm entitled as Delayed Shortest Normal Processing Time (DSNPT), matching the lower bound proposed by us, and hence DSNPT is optimal.

Details DOI

TCS Journal 2021 Journal Article

Bicriteria algorithms to balance coverage and cost in team formation under online model

Yijing Wang
Dachuan Xu
Donglei Du
Ran Ma

In this work, we investigate online bicriteria algorithms that consider both coverage and cost in the team formation problem, which selects a set of experts with the objective of maximizing the difference of two set functions f − ℓ, where function f is non-negative normalized monotone approximately submodular, and function ℓ is non-negative linear. By exploiting the problem's combinatorial structure, we present three bicriteria algorithms along with their corresponding competitive analysis. The first two algorithms handle the cases where function f is γ-weakly submodular, and strictly γ-weakly submodular, respectively. The last algorithm is more general by integrating the first two with extra parameters introduced.

Details DOI

TCS Journal 2014 Journal Article

Primary–secondary bicriteria scheduling on identical machines to minimize the total completion time of all jobs and the maximum T-time of all machines

Long Wan
Ran Ma
Jinjiang Yuan

In this paper, we study a new primary–secondary bicriteria scheduling problem on identical machines. The primary objective is to minimize the total completion time of all jobs and the secondary objective is to minimize the maximum T-time of all machines, where the T-time of a machine is defined as the total completion time of the jobs scheduled on the machine. The problem is to find a non-preemptive schedule of minimizing the secondary objective subject to the constraint that the primary objective is minimized. It is implied in the literature that the problem is ordinarily NP-hard if the number of machines is fixed, and strongly NP-hard if the number of machines is a part of input. When the number of machines is fixed, we present a pseudo-polynomial-time algorithm and a fully polynomial-time approximation scheme. Then we analyze the classic algorithm SPT which schedules jobs to machines greedily in the order of non-decreasing processing times. We show that the worst-case ratio of SPT is of at most 11/6 and at least 5/3. Furthermore, we present another algorithm, called RSPT, with the worst-case ratio of at most 3/2 and at least 11/9.

Details DOI