The end-to-end automated design of machine learning (ML) pipelines significantly reduces the workload for data scientists and democratizes ML for non-experts. Evolutionary algorithm (EA)-based automated ML (AutoML) systems, a prominent category of AutoML, often face inefficiencies due to the costly fitness evaluation of candidate ML pipelines. Although surrogate models have been employed to approximate the true performance of pipelines more quickly, a key challenge remains in effectively bridging the semantic gap between the heterogeneous features of datasets and pipelines. To address this issue, we propose ADELA, a novel accompanying surrogate-based optimization strategy that accelerates EA-based AutoML while retaining the performance of the resulting pipelines. ADELA operates in two phases: Offline, leveraging a high-quality curated pipeline corpus to meta-learn an accompanying surrogate model; and Online, selecting the accompanying pipeline and using the learned model to predict the performance of evaluation pipelines instead of executing them. The accompanying mechanism effectively mitigates the semantic gap between datasets and pipelines, enabling ADELA to reduce computation times by an average of 73.66% while retaining 98.78% of the final pipeline performance, as demonstrated in extensive experimental evaluations.

The pursuit of personalized outfit recommendation, tailored to individual user preferences, has emerged as a central focus in research. Despite the propositions of various outfit recommendation methods, the modeling of fashion styles within outfits, a pivotal criterion for user selection, has been largely overlooked. Fashion styles are intricately embedded in multi-modal information, such as images and text. Unfortunately, current outfit recommendation methods predominantly focus on a single modality or employ simplistic fusion techniques, thus failing to capture the intricate interplay between different fashion data modalities, resulting in an inability to model fashion styles effectively. In this work, we devise a novel Personalized Outfit Recommender with Style-Guided Multi-Modal Feature Fusion scheme, denoted as POSM, aiming to maximize the utilization of multi-modal features in outfit recommendation. In particular, this scheme consists of three key components: modality-aware outfit modeling, feature separation network, and style mapping component. Extensive experiments conducted on the benchmark datasets validate the superiority of POSM over state-of-the-art methods.