Federated Unlearning (FU) has emerged as a promising paradigm for effectively removing the influence of specific data of clients from the global model in federated learning. It can further enhance personal data privacy for individual clients and eliminate the impact of malicious attacks like poisoning. Due to these benefits, many FU methods have been analyzed and proposed. Yet, they largely overlook external threats against FU systems, such as gradient inversion attacks that reconstruct client data from shared gradients, posing serious privacy risks to other participating clients. Motivated by this, we propose DIARY, a Differential prIvacy IntegrAted fedeRated recoverY framework to address these dual threats. DIARY presents a privacy budget allocation method, whose insight lies in adaptively assigning appropriate privacy budgets to various training and recovery phases to fully utilize the global privacy budget of each client, balancing the trade-off between privacy and utility. Furthermore, DIARY introduces a novel Federated noise-Immune aNomaly Detection (FIND) module. The deep integration of FIND with two-level selective storage and model rollback mechanisms contributes to model recovery, while significantly reducing the associated overhead. Finally, both rigorous theoretical analysis and extensive simulations compared with state-of-the-art methods are conducted to validate the effectiveness of DIARY.

The 532-nm green laser light commonly used for airborne laser bathymetry (ALB) can penetrate clear shallow water but is sensitive to turbidity, which could lead to water surface uncertainty. In this study, water surface uncertainty was quantitatively assessed using a photon-counting LiDAR (PCL) with high receiver sensitivity to analyze the effect of turbidity. The qualitative results showed that the water surface heights are generally underestimated, and the surface detection accuracy in turbid water is superior to that in clear water. These findings were confirmed by statistical analysis of representative data in quantitative empirical experiments. The diffuse attenuation coefficient as a metric of water turbidity ranged from 0.14 to 4.80$\text{m}^{-1}$for clear to turbid water. The corresponding underestimated deviation ranged from 0.37 to 0.08 m, and the root-mean-square error (RMSE) was ranged from 0.39 to 0.06 m. In addition, the radiative transfer mechanism underlying the underestimation of water surface heights at different levels as water turbidity varies was determined by comparing the simulated and measured results. On the one hand, there is a high exponential relationship between the underestimation deviation and the diffuse attenuation coefficient when considering only the water optical properties. On the other hand, the presence of direct reflection component from the surface actually has an inhibiting effect on the underestimation. The present study provides reliable evidence for further understanding the interaction of green lasers with the air–water interfaces.

In recent years, various coral reef retrieval methods have experienced considerable progress due to a variety of observational instruments and innovative parameter calculation techniques. However, these labor-intensive methods are deficient in handling high-precision remote sensing mapping of coral reef benthic environments, facing challenges, including enhancing the robustness of scaled coral reef retrieval against varying water depths and complex water column conditions. To overcome these limitations, we propose a novel method for coral reef benthic mapping. Our method primarily consists of two central components: water depth extraction and coral reef information acquisition. Accurate bathymetry is critical for coral reef remote sensing inversion. To obtain more precise water depth, we propose the Bathymetry Transformer model. Our Bathymetry Transformer model aggregates vast amounts of active and passive remote sensing data, generates accurate bathymetry (with a 0.375-m RMSE, ranging from 0- to 12-m water depth) that eliminates the need for in situ examinations, and maintains a harmonious balance between greater bathymetry precision and finer spatial resolution. Based on this, water column corrections are then applied, and the results derived from various active and passive remote sensing processes, along with their respective band calculation outcomes, are fed into the proposed coral reef Transformer (CR Transformer) model to generate high-accuracy coral reef benthic mapping results. Copious experimental outcomes affirm that our CR Transformer surpasses current state-of-the-art (SOTA) methods in computational efficiency and results accuracy. Impressively, the CR Transformer achieves a notable mean intersection over union (mIoU) of 91.25% and an accuracy of 95.71% on the validation dataset.

Patents are very important intangible assets that protect firm technologies and maintain market competitiveness. Thus, patent evaluation is critical for firm business strategy and innovation management. Currently patent evaluation mostly relies on some meta information of patents, such as number of forward/backward citations and number of claims. In this paper, we propose to identify patent technological trends, which carries information about technology evolution and trajectories among patents, to enable more effective and precise patent evaluation. We explore features to capture both the value of trends and the quality of patents within a trend, and perform patent evaluation to validate the extracted trends and features using patents in the United States Patent and Trademark Office (USPTO) dataset. Experimental results demonstrate that the identified technological trends are able to capture patent value precisely. With the proposed trend related features extracted from our identified trends, we can improve patent evaluation performance significantly over the baseline using conventional features.