This study addresses the limitations of existing dynamic pseudo-labeling (DPL) techniques, which often utilize static or dynamic thresholds for confident sample selection. The existing methods fail to capture the non-linear relationship between task accuracy and model confidence, particularly in the context of overconfidence. This can limit the model’s learning opportunities for high entropy samples that significantly influence a model’s generalization ability. To solve this, we propose a novel gradient pass-based DPL technique that incorporates the high-entropy samples, which are typically overlooked. Our approach introduces two classifiers–low gradient pass (LGP) and high gradient pass (HGP)–to derive over- and under-confident dynamic thresholds that indicate the class-wise overconfidence acceleration, respectively. By combining the under- and overconfident states from the GP classifiers, we create a more adaptive and accurate PL method. Our main contributions highlight the importance of considering both low and high-confidence samples in enhancing the model’s robustness and generalization for improved PL performance.

Source-free domain adaptation (SFDA) aims to adapt a pre-trained source model to an unlabeled target domain without requiring labeled source data. In a self supervised setting, relying on pseudo labels on target domain samples facilitates the domain adaptation performance providing strong supervision. However, a critical problem of this approach is the inherent instability of the pre-trained source model in the target domain, leading to unreliable pseudo labels for the target domain data. To tackle this, we propose a novel Dual-perspective pseudo labeling strategy that jointly leverages a task-specific perspective and a domain-invariant perspective, assigning pseudo labels only to target samples on which the target model’s predictions and CLIP’s predictions agree. To further enhance representation learning without introducing noisy supervision, we apply consistency training to uncertain samples. Additionally, we introduce a Tsallis mutual information(TMI)-based vision optimization strategy guided by an Uncertainty-based adaptation index (UAI), which dynamically modulates entropy sensitivity based on the model’s adaptation uncertainty. The UAI-based training paradigm enables stable and adaptive domain alignment by effectively balancing exploration and exploitation processes during the optimization process. Our proposed method achieves state-of-the-art performance on domain adaptation benchmark datasets, improving adaptation accuracy by 1.6% on Office-Home, 1.4% on VisDA-C, and 2.9% on DomainNet-126, demonstrating its effectiveness in SFDA. The code is publicly available at https://github.com/l3umblee/duet-sfda.

While existing supervised deep learning-based lane detection methods achieve exceptional detection performance, constructing a large real-world dataset with labels is a cost-intensive task. Therefore, we propose a novel sim-to-real unsupervised domain adaptation method specialized in lane detection. In this paper, we present a disentangled feature alignment approach that performs selective adaptation for lane and background features. By performing the disentangled prototype-based local and global feature alignments, we solve the negative transfer problem of an existing adversarial feature alignment-based domain adaptation for lane detection. In addition, the lane detection task requires accurate localization of the lanes even in occluded parts. Therefore, we adopt a consistency regularization strategy using masked images to improve the detection accuracy in the occluded area. Additionally, we introduce an adaptive resolution adjustment of the masked-out patch based on training maturity for context learning. By applying a simple framework, we achieve superior lane detection accuracy with lower false positives and false negatives in the target domain compared to conventional methods. Extensive experiments demonstrate the superiority of the proposed method.

Abstract This study proposes a new visualization method and index for collection evaluation. Specifically, it develops a network‐based mapping technique and a user‐focused Hirsch index (user‐side h‐index) given the lack of previous studies on collection evaluation methods that have used the h‐index. A user‐side h‐index is developed and compared with previous indices (use factor, difference of percentages, collection‐side h‐index) that represent the strengths of the subject classes of a library collection. The mapping procedure includes the subject‐usage profiling of 63 subject classes and collection‐usage map generations through the pathfinder network algorithm. Cluster analyses are then conducted upon the pathfinder network to generate 5 large and 14 small clusters. The nodes represent the strengths of the subject‐class usages reflected by the user‐side h‐index. The user‐side h‐index was found to have advantages (e.g., better demonstrating the real utility of each subject class) over the other indices. It also can more clearly distinguish the strengths between the subject classes than can collection‐side h‐index. These results may help to identify actual usage and strengths of subject classes in library collections through visualized maps. This may be a useful rationale for the establishment of the collection‐development plan.

Statistical association measures have been widely applied in information retrieval research, usually employing a clustering of documents or terms on the basis of their relationships. Applications of the association measures for term clustering include automatic thesaurus construction and query expansion. This research evaluates the similarity of six association measures by comparing the relationship and behavior they demonstrate in various analyses of a test corpus. Analysis techniques include comparisons of highly ranked term pairs and term clusters, analyses of the correlation among the association measures using Pearson's correlation coefficient and MDS mapping, and an analysis of the impact of a term frequency on the association values by means of z-score. The major findings of the study are as follows: First, the most similar association measures are mutual information and Yule's coefficient of colligation Y, whereas cosine and Jaccard coefficients, as well as χ2 statistic and likelihood ratio, demonstrate quite similar behavior for terms with high frequency. Second, among all the measures, the χ2 statistic is the least affected by the frequency of terms. Third, although cosine and Jaccard coefficients tend to emphasize high frequency terms, mutual information and Yule's Y seem to overestimate rare terms.