Zijian Chen, Xueguang Ma, Shengyao Zhuang, Ping Nie, Kai Zou, Sahel Sharifymoghaddam, Andrew Liu, Joshua Green, Kshama Patel, Ruoxi Meng, Mingyi Su, Yanxi Li, Haoran Hong, Xinyu Shi, Xuye Liu, Hosna Oyarhoseini, Nandan Thakur, Crystina Zhang, Luyu Gao, Wenhu Chen, Jimmy Lin. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Quality‐aware service integration in cloud computing information infrastructures typically refers to selecting a suitable subset of services out of those available in order to fulfill a user’s request, considering multiple quality of service (QoS) metrics like response time, cost, availability, and reliability constraints. Since this optimization problem is NP‐complete and possesses a large search space, it remains an active topic of cloud computing research. Most metaheuristic techniques developed to solve this problem work under assumptions of relatively static‐QoS conditions and require continuous service monitoring in order to function. However, modern cloud environments often exhibit highly dynamic workloads and changing resource availabilities, leading to varying QoS characteristics. This work proposes a hybrid deep learning and metaheuristic‐based framework for QoS‐aware cloud service integration. It synergizes a deep learning QoS prediction model and the elephant herding optimization (EHO) algorithm for search and delivers near‐optimal service integration decisions. EHO was chosen over the most recent/metahybrid evolutionary algorithms due to its empirical efficiency and effectiveness in tackling the large‐scale QoS‐aware service integration problem. The proposed method consistently outperforms existing baselines like genetic algorithm (GA), particle swarm optimization (PSO), and stand‐alone EHO in terms of QoS metrics measured on a simulated cloud platform with publicly available service datasets. The average response time of the selected services under the proposed method is between 20% and 45% lower, service availability is always above 99.5%, the service cost can be up to 40% lower, and reliability can be increased by about 2%–4%. In addition to these performance gains, the benefits of the proposed framework include decoupling service integration decisions from real‐time monitoring of cloud resources, due to QoS prediction. Specifically, a deep learning–based forecasting model is employed to quickly adapt to dynamic/cloud workloads and produce reliable predictions of QoS values even when the underlying cloud infrastructure is changing rapidly. This allows for applying the proposed framework to challenging large‐scale and complicated environments like IoT infrastructures, smart cities, and industrial cloud platforms, which need to take scalability, heterogeneity of service compositions, and limited resource capabilities of edge devices into account. This predictive model also offers a valuable tool for designing future cloud orchestration systems, integrating seamlessly with current real‐time cloud management platforms. The limitation of this work is the dependency on acquiring enough historical data to train the deep learning model. One possible way to address this is through transfer learning or data augmentation techniques.

Motivated by the success of fractional pixel motion in video coding, we explore the design of motion estimation with fractional-voxel resolution for compression of color attributes of dynamic 3D point clouds. Our proposed block-based fractional-voxel motion estimation scheme takes into account the fundamental differences between point clouds and videos, i.e., the irregularity of the distribution of voxels within a frame and across frames. We show that motion compensation can benefit from the higher resolution reference and more accurate displacements provided by fractional precision. Our proposed scheme significantly outperforms comparable methods that only use integer motion. The proposed scheme can be combined with and add sizeable gains to state-of-the-art systems that use transforms such as Region Adaptive Graph Fourier Transform and Region Adaptive Haar Transform.

In point cloud compression, exploiting temporal redundancy for inter predictive coding is challenging because of the irregular geometry. This paper proposes an efficient block-based inter-coding scheme for color attribute compression. The scheme includes integer-precision motion estimation and an adaptive graph based in-loop filtering scheme for improved attribute prediction. The proposed block-based motion estimation scheme consists of an initial motion search that exploits geometric and color attributes, followed by a motion refinement that only minimizes color prediction error. To further improve color prediction, we propose a vertex-domain low-pass graph filtering scheme that can adaptively remove noise from predictors computed from motion estimation with different accuracy. Our experiments demonstrate significant coding gain over state-of-the-art coding methods.

Abstract With a core purpose of helping users to understand the context, a water interface provides possibility for enhancing user experience in interaction process. Starting from analyzing existing water‐mediated interaction approaches, we proposed a water‐mediated interaction design model and a corresponding user experience model, aiming to eliminate the boundary between users and the context with water as the medium. According to the proposed model, we implemented a water‐mediated interaction system Landscape Rippling, with the painting “A Panorama of Rivers and Mountains” as its context. Ultimately, user experience tests of the interaction system demonstrate the effectiveness of this water‐mediated interaction design model.

An unfocused plenoptic light field (LF) camera places an array of microlenses in front of an image sensor in order to separately capture different directional rays arriving at an image pixel. Using a conventional Bayer pattern, data captured at each pixel is a single color component (R, G or B). The sensed data then undergoes demosaicking (interpolation of RGB components per pixel) and conversion to an array of sub-aperture images (SAIs). In this paper, we propose a new LF image coding scheme based on graph lifting transform (GLT), where the acquired sensor data are coded in the original captured form without pre-processing. Specifically, we directly map raw sensed color data to the SAIs, resulting in sparsely distributed color pixels on 2D grids, and perform demosaicking at the receiver after decoding. To exploit spatial correlation among the sparse pixels, we propose a novel intra-prediction scheme, where the prediction kernel is determined according to the local gradient estimated from already coded neighboring pixel blocks. We then connect the pixels by forming a graph, modeling the prediction residuals statistically as a Gaussian Markov Random Field (GMRF). The optimal edge weights are computed via a graph learning method using a set of training SAIs. The residual data is encoded via low-complexity GLT. Experiments show that at high PSNRs-important for archiving and instant storage scenarios-our method outperformed significantly a conventional light field image coding scheme with demosaicking followed by High Efficiency Video Coding (HEVC).