Yilun Zhu, Yuan Zhuang, Nikhita Vedula, Dushyanta Dhyani, Shaoyuan Xu, Mohsen Bayati, Bryan Wang, Shervin Malmasi. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Recently, multi-objective optimization (MOO) has gained attention for its broad applications in ML, operations research, and engineering. However, MOO algorithm design remains in its infancy and many existing MOO methods suffer from unsatisfactory convergence rate and sample complexity performance. To address this challenge, in this paper, we propose an algorithm called STIMULUS (**st**ochastic path-**i**ntegrated **mul**ti-gradient rec**u**rsive e**s**timator), a new and robust approach for solving MOO problems. Different from the traditional methods, STIMULUS introduces a simple yet powerful recursive framework for updating stochastic gradient estimates to improve convergence performance with low sample complexity. In addition, we introduce an enhanced version of \algns, termed \algmns, which incorporates a momentum term to further expedite convergence. We establish $\mathcal{O}(1/T)$ convergence rates of the proposed methods for non-convex settings and $\mathcal{O}(\exp{-\mu T})$ for strongly convex settings, where $T$ is the total number of iteration rounds. Additionally, we achieve the state-of-the-art $O\left(n+\sqrt{n}\epsilon^{-1}\right)$ sample complexities for non-convex settings and $\mathcal{O}\left(n+ \sqrt{n} \ln ({\mu/\epsilon})\right)$ for strongly convex settings, where $\epsilon>0$ is a desired stationarity error. Moreover, to alleviate the periodic full gradient evaluation requirement in STIMULUS and STIMULUS-M, we further propose enhanced versions with adaptive batching called STIMULUS$^+$/ STIMULUS-M$^+$ and provide their theoretical analysis.

For image restoration, methods leveraging priors from generative models have been proposed and demonstrated a promising capacity to robustly restore photorealistic and high-quality results. However, these methods are susceptible to semantic ambiguity, particularly with images that have obviously correct semantics such as facial images. In this paper, we propose a semantic-aware latent space exploration method for image restoration (SAIR). By explicitly modeling semantics information from a given reference image, SAIR is able to reliably restore severely degraded images not only to high-resolution and highly realistic looks but also to correct semantics. Quantitative and qualitative experiments collectively demonstrate the superior performance of the proposed SAIR. Our code is available in https://github.com/Liamkuo/SAIR.

On shopping websites, product images of low quality negatively affect customer experience. Although there are plenty of work in detecting images with different defects, few efforts have been dedicated to correct those defects at scale. A major challenge is that there are thousands of product types and each has specific defects, therefore building defect specific models is unscalable. In this paper, we propose a unified Image-to-Image (I2I) translation model to correct multiple defects across different product types. Our model leverages an attention mechanism which hierarchically incorporates high-level defect groups and specific defect types to guide the network to focus on defect-related image regions. Evaluated on eight public datasets, our model reduces the Frechet Inception Distance (FID) by 24.6% in average compared with MoNCE, the state-of-the-art I2I method. Another practical challenge on shopping websites is the lack of high quality paired images. We extend our model to be semi-paired by leveraging both paired and unpaired data. Tested on a shopping website dataset to correct three image defects, our model reduces (FID) by 63.2% in average compared with WS-I2I, the state-of-the art semi-paired I2I method.

Semi-supervised learning (SSL) has become a crucial approach in deep learning as a way to address the challenge of limited labeled data. The success of deep neural networks heavily relies on the availability of large-scale high-quality labeled data. However, the process of data labeling is time-consuming and unscalable, leading to shortages in labeled data. SSL aims to tackle this problem by leveraging additional unlabeled data in the training process. One of the popular SSL algorithms, FixMatch [1], trains identical weight-sharing teacher and student networks simultaneously using a siamese neural network (SNN). However, it is prone to performance degradation when the pseudo labels are heavily noisy in the early training stage. We present KD-FixMatch, a novel SSL algorithm that addresses the limitations of FixMatch by incorporating knowledge distillation. The algorithm utilizes a combination of sequential and simultaneous training of SNNs to enhance performance and reduce performance degradation. Firstly, an outer SNN is trained using labeled and unlabeled data. After that, the network of the well-trained outer SNN generates pseudo labels for the unlabeled data, from which a subset of unlabeled data with trusted pseudo labels is then carefully created through high-confidence sampling and deep embedding clustering. Finally, an inner SNN is trained with the labeled data, the unlabeled data, and the subset of unlabeled data with trusted pseudo labels. Experiments on four public data sets demonstrate that KD-FixMatch outperforms FixMatch in all cases. Our results indicate that KD-FixMatch has a better training starting point that leads to improved model performance compared to FixMatch.

Multi-modal learning with both text and images benefits multiple applications, such as attribute extraction for e-commerce products. In this paper, we propose Cross-Modality Attention Contrastive Language-Image Pre-training (CMA-CLIP), a new multi-modal architecture to jointly learn the fine-grained inter-modality relationship. It fuses CLIP with a sequence-wise attention module and a modality-wise attention module. The network uses CLIP to bridge the inter-modality gap at the global level, and uses the sequence-wise attention module to capture the fine-grained alignment between text and images. Besides, it leverages a modality-wise attention module to learn the relevance of each modality to downstream tasks, making the network robust against irrelevant modalities. CMA-CLIP outperforms the state-of-the-art method on Fashion-Gen by 5.5% in accuracy, achieves competitive performance on Food101 and performance on par with the state-of-the-art method on MM-IMDb. We also demonstrate CMA-CLIP’s robustness against irrelevant modalities on an Amazon dataset for the task of product attribute extraction.