TCL CSOT

Lu Wang, Xiaofu Chang, Shuang Li, Yunfei Chu, Hui Li, Wei Zhang, Xiao-Feng He, Le Song, Jing-Ren Zhou, Hong-Xia Yang

Dynamic graph modeling has recently attracted much attention due to its extensive applications in many real-world scenarios, such as recommendation systems, financial transactions, and social networks. Although many works have been proposed for dynamic graph modeling in recent years, effective and scalable models are yet to be developed. In this paper, we propose a novel graph neural network approach, called TCL, which deals with the dynamically-evolving graph in a continuous-time fashion and enables effective dynamic node representation learning that captures both the temporal and topology information. Technically, our model contains three novel aspects. First, we generalize the vanilla Transformer to temporal graph learning scenarios and design a graph-topology-aware transformer. Secondly, on top of the proposed graph transformer, we introduce a two-stream encoder that separately extracts representations from temporal neighborhoods associated with the two interaction nodes and then utilizes a co-attentional transformer to model inter-dependencies at a semantic level. Lastly, we are inspired by the recently developed contrastive learning and propose to optimize our model by maximizing mutual information (MI) between the predictive representations of two future interaction nodes. Benefiting from this, our dynamic representations can preserve high-level (or global) semantics about interactions and thus is robust to noisy interactions. To the best of our knowledge, this is the first attempt to apply contrastive learning to representation learning on dynamic graphs. We evaluate our model on four benchmark datasets for interaction prediction and experiment results demonstrate the superiority of our model.

Wanxing Chang, Ye Shi, Jing-Ya Wang

Learning with noisy labels (LNL) poses a significant challenge in training a well-generalized model while avoiding overfitting to corrupted labels. Recent advances have achieved impressive performance by identifying clean labels and correcting corrupted labels for training. However, the current approaches rely heavily on the model's predictions and evaluate each sample independently without considering either the global and local structure of the sample distribution. These limitations typically result in a suboptimal solution for the identification and correction processes, which eventually leads to models overfitting to incorrect labels. In this paper, we propose a novel optimal transport (OT) formulation, called Curriculum and Structure-aware Optimal Transport (CSOT). CSOT concurrently considers the inter- and intra-distribution structure of the samples to construct a robust denoising and relabeling allocator. During the training process, the allocator incrementally assigns reliable labels to a fraction of the samples with the highest confidence. These labels have both global discriminability and local coherence. Notably, CSOT is a new OT formulation with a nonconvex objective function and curriculum constraints, so it is not directly compatible with classical OT solvers. Here, we develop a lightspeed computational method that involves a scaling iteration within a generalized conditional gradient framework to solve CSOT efficiently. Extensive experiments demonstrate the superiority of our method over the current state-of-the-arts in LNL. Code is available at https://github.com/changwxx/CSOT-for-LNL.

Zhenhe Zhang, Jun Zheng, Guchuan Zhu

This paper presents a study on event-triggered power tracking control of heterogeneous thermostatically controlled load (TCL) populations. The developed schemes are based on continuous-time tracking control of TCL populations of which the aggregated dynamics are described by coupled Fokker-Planck equations. Two event-triggering mechanisms, namely static and dynamic event-triggered control strategies, are proposed, which can guarantee the input-to-state practical stability (ISpS) of the tracking error dynamics while excluding Zeno phenomenon. A simulation study is conducted, and the obtained results show that the developed control strategies can significantly reduce the communication burden while still offering a satisfactory control performance.

Ting Shi, Xue-Bin Chen, Xiang-Long Li, Guangyao Li, Bai-Xiang Han, Zhaosong Liu, Jin-Chuan Li, Wei-Ran Cao

TCL CSOT utilized inkjet printing technology to produce impressive medium and large‐size OLED products in the previous. In this work, high‐resolution notebook equipped with IGZO backplane is shown. This inkjet printing notebook offers 240 PPI along with good optical performance. Also, breakthrough in high‐resolution printing has been achieved, 274 PPI panel has been successfully printed. It is demonstrated that inkjet printing technology would adapt the trend of display products with various sizes toward higher resolution.

Cheng-Lei Nie, Jing-Chun Guo, Li-Ying Zhang, Yan-Ying Du, Xiang-Zhen Kong, Wei-Ran Cao

This work indicates the latest progress of 27‐inch 4K inkjet printing OLED display that currently developed by TCL CSOT. To enhance the printed layers uniformity that significantly affects panel performance, the microcavity structure has been optimized using a new device design featuring with thickened anode. After the combined evaluation of device performance, process capability and mass‐produced feasibility, thickened IZO is ultimately chosen as the anode material.