In the real world, where information is abundant and diverse across different modalities, understanding and utilizing various data types to improve retrieval systems is a key focus of research. Multimodal composite retrieval integrates diverse modalities such as text, image and audio to provide more accurate, personalized, and contextually relevant results. However, alongside retrieval, multimodal composite editing plays a crucial role in enabling users to refine or modify retrieved content through intuitive interactions, which enhances the overall effectiveness of multimodal systems. The task of multimodal composite editing is becoming increasingly critical due to its applications in various domains, including creative industries, education, and user-driven content modification. A comprehensive evaluation and usage guide is needed to fully assess its capabilities and limitations, since it complements and extends the functionalities provided by multimodal retrieval systems. To facilitate a deeper understanding of this promising direction, this survey explores multimodal composite editing and retrieval in depth, covering image-text composite editing, image-text composite retrieval, and other multimodal composite retrieval. In this survey, we systematically organize the application scenarios, methods, benchmarks, experiments, and future directions. Multimodal learning has gained significant popularity in the era of large AI models, as demonstrated by the growing number of surveys in multimodal learning and vision-language models with Transformers. To the best of our knowledge, this survey is the first comprehensive review of the literature on multimodal composite retrieval, which is a timely complement of multimodal fusion to existing reviews. Moreover, this paper bridges the gap between large model architectures and their applications in both retrieval and editing tasks, highlighting their intertwined roles in advancing multimodal systems.

Labeled data scarcity of an interested domain is often a serious problem in machine learning. Leveraging the labeled data from other semantic-related yet co-variate shifted source domain to facilitate the interested domain is a consensus. In order to solve the domain shift between domains and reduce the learning ambiguity, unsupervised domain adaptation (UDA) greatly promotes the transferability of model parameters. However, the dilemma of over-fitting (negative transfer) and under-fitting (under-adaptation) is always an overlooked challenge and potential risk. In this paper, we rethink the shallow learning paradigm and this intractable over/under-fitting problem, and propose a safer UDA model, coined as Bilateral Co-Transfer (BCT), which is essentially beyond previous well-known unilateral transfer. With bilateral co-transfer between domains, the risk of over/under-fitting is therefore largely reduced. Technically, the proposed BCT is a symmetrical structure, with joint distribution discrepancy (JDD) modeled for domain alignment and category discrimination. Specifically, a symmetrical bilateral transfer (SBT) loss between source and target domains is proposed under the philosophy of mutual checks and balances. First, each target sample is represented by source samples with low-rankness constraint in a common subspace, such that the most informative and transferable source data can be used to alleviate negative transfer. Second, each source sample is symmetrically and sparsely represented by target samples, such that the most reliable target samples can be exploited to tackle under-adaptation. Experiments on various benchmarks show that our BCT outperforms many previous outstanding work.