Abstract In the current realms of scientific research and commercial applications, the risk inference of regulatory violations by publicly listed enterprises has attracted considerable attention. However, there are some problems in the existing research on the deduction and prediction of violation risk of listed enterprises, such as the lack of analysis of the causal logic association between violation events, the low interpretability and effectiveness of the deduction and the lack of training data. To solve these problems, we propose a framework for enterprise violation risk deduction based on generative AI and event evolution graphs. First, the generative AI technology was used to generate a new text summary of the lengthy and complex enterprise violation announcement to realize a concise overview of the violation matters. Second, by fine‐tuning the generative AI model, an event entity and causality extraction framework based on automated data augmentation are proposed, and the UIE (Unified Structure Generation for Universal Information Extraction) event entity extraction model is used to create the event entity extraction for listed enterprises ‘violations. Then, a causality extraction model CDDP‐GAT (Event Causality Extraction Based on Chinese Dictionary and Dependency Parsing of GAT) is proposed. This model aims to identify and analyse the causal links between corporate breaches, thereby deepening the understanding of the event logic. Then, the merger of similar events was realized, and the causal correlation weights between enterprise violation‐related events were evaluated. Finally, the listed enterprise's violation risk event evolution graph was constructed, and the enterprise violation risk deduction was carried out to form an expert system of financial violations. The deduction results show that the method can effectively reveal signs of enterprise violations and adverse consequences.

We introduceTreeStructor, a novel approach for isolating and reconstructing forest trees. The key novelty is a deep neural model that uses neural ranking to assign pre-generated connectable 3D geometries to a point cloud.TreeStructoris trained on a large set of synthetically generated point clouds. The input to our method is a forest point cloud (FPC) that we first decompose into point clouds that approximately represent trees (TPC) and then into point clouds that represent their parts (PPC). We use a point cloud encoder-decoder to compute embedding vectors that retrieve the best-fitting surface mesh for eachPPCfrom a large set of predefined branch parts. Finally, the retrieved meshes are connected and oriented to obtain individual surface meshes of all trees represented by theFPC. We qualitatively and quantitatively validate that our method can reconstruct forest trees with unprecedented accuracy and visual fidelity.TreeStructoroutperforms the state-of-the-art reconstruction method for around 6% on quantitative metrics and 12% less error compared with QSM on low-quality scanned data. The code and data are available at https://lewkesy.github.io/TreeStructor/.

Presents corrections to the paper, (Errata to “TreeStructor: Forest Reconstruction With Neural Ranking”).

Training data in federated learning (FL) frameworks can have label noise, since they must be stored and annotated on clients' devices. If trained over such corrupted data, the models learn the wrong knowledge of label noise, which highly degrades their performance. Although several FL schemes are designed to combat label noise, they suffer performance degradation when the clients' devices only have limited local training samples. To this end, a new scheme called federated label-noise learning (FedLNL) is developed in this paper. The key problem of FedLNL is how to estimate a noise transition matrix (NTM) accurately in the case of limited local training samples. If a gradient-based update method is used to update the local NTM on each client's device, it can generate too large gradients for the local NTM, causing a high estimation error of the local NTM. To tackle this issue, an alternating update method for the local NTM and the local classifier is designed in FedLNL, where the local NTM is updated by a Bayesian inference-based update method. Such an alternating update method makes the loss function of existing NTM-based schemes not applicable to FedLNL. To enable federated optimization of FedLNL, a new regularizer on the parameters of the classifier called local diversity product regularizer is designed for the loss function of FedLNL. The results show that FedLNL improves the test accuracy of a trained model by up to 25.98%, compared with the state-of-the-art FL schemes that tackle label-noise issues.

It is prevalent for a mobile edge device to conduct local inference using a compact machine learning model, which achieves lower latency and less compromise of data privacy as compared to cloud-based inference. To work in a new environment, the compact model needs to be adapted to the target data from the environment so as to maintain a high inference accuracy. However, directly applying domain adaptation to the compact model leads to a low inference accuracy. Hence, a scheme called memory-efficient collaborative domain adaptation (MEC-DA) is developed in this paper to boost the compact model's inference accuracy on the target data while preserving data privacy. It first deploys a large model to the mobile edge devices where domain adaptation is conducted to adapt the large model to the target data. This process requires training of the large model, which causes high memory consumption. A new method called lite residual hypothesis transfer (LRHT) is thus designed to achieve memory-efficient domain adaptation. The knowledge of the large model is then transferred to the compact model via knowledge distillation. To prevent the compact model from forgetting the knowledge of the source data, a collaborative knowledge distillation (Co-KD) method is developed to unify the source data on the server and the target data on an edge device to update the compact model. MEC-DA can protect data privacy and handle user mobility properly via secure aggregation and user selection, respectively. Extensive experiments on several tasks of object recognition show that MEC-DA improves the inference accuracy by up to$12.5\%$, as compared to the state-of-the-art schemes.

A federated kernel k -means (FedKKM) algorithm is developed in this article to conduct distributed clustering with low memory consumption on user devices. In FedKKM, a federated eigenvector approximation (FEA) algorithm is designed to iteratively determine the low-dimensional approximate vectors of the transformed feature vectors, using only low-dimensional random feature vectors. To maintain high communication efficiency in each iteration of FEA, a communication-efficient Lanczos algorithm (CELA) is further designed in FEA to reduce the communication cost. Based on the low-dimensional approximate vectors, the clustering result is obtained by leveraging a distributed linear k -means algorithm. A theoretical analysis shows that: 1) FEA has a convergence rate of O(1/T) , where T is the number of iterations; 2) the scalability of FedKKM is not affected by the dataset size since the communication cost of FedKKM is independent of the number of users' data; and 3) FedKKM is a (1+ϵ) approximation algorithm. The experimental results show that FedKKM achieves the comparable clustering quality to that of a centralized kernel k -means. Compared with state-of-the-art schemes, FedKKM reduces the memory consumption on user devices by up to 94% and also reduces the communication cost by more than 40%.

A federated learning (FL) scheme (denoted as Fed-KSVM) is designed to train kernel support vector machines (SVMs) over multiple edge devices with low memory consumption. To decompose the training process of kernel SVM, each edge device first constructs high-dimensional random feature vectors of its local data, and then trains a local SVM model over the random feature vectors. To reduce the memory consumption on each edge device, the optimization problem of the local model is divided into several subproblems. Each subproblem only optimizes a subset of the model parameters over a block of random feature vectors with a low dimension. To achieve the same optimal solution to the original optimization problem, an incremental learning algorithm called block boosting is designed to solve these subproblems sequentially. After training of the local models, the central server constructs a global SVM model by averaging the model parameters of these local models. Fed-KSVM only increases the iterations of training the local SVM models to save the memory consumption, while the communication rounds between the edge devices and the central server are not affected. Theoretical analysis shows that the kernel SVM model trained by Fed-KSVM converges to the optimal model with a linear convergence rate. Because of such a fast convergence rate, Fed-KSVM reduces the communication cost during training by up to 99% compared with the centralized training method. The experimental results also show that Fed-KSVM reduces the memory consumption on the edge devices by nearly 90% while achieving the highest test accuracy, compared with the state-of-the-art schemes.

In this article, we propose DeepTree, a novel method for modeling trees based on learning developmental rules for branching structures instead of manually defining them. We call our deep neural model "situated latent" because its behavior is determined by the intrinsic state -encoded as a latent space of a deep neural model- and by the extrinsic (environmental) data that is "situated" as the location in the 3D space and on the tree structure. We use a neural network pipeline to train a situated latent space that allows us to locally predict branch growth only based on a single node in the branch graph of a tree model. We use this representation to progressively develop new branch nodes, thereby mimicking the growth process of trees. Starting from a root node, a tree is generated by iteratively querying the neural network on the newly added nodes resulting in the branching structure of the whole tree. Our method enables generating a wide variety of tree shapes without the need to define intricate parameters that control their growth and behavior. Furthermore, we show that the situated latents can also be used to encode the environmental response of tree models, e.g., when trees grow next to obstacles. We validate the effectiveness of our method by measuring the similarity of our tree models and by procedurally generated ones based on a number of established metrics for tree form.

Mo Yu, Jiangnan Li, Shunyu Yao, Wenjie Pang, Xiaochen Zhou, Zhou Xiao, Fandong Meng, Jie Zhou. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Federated learning (FL) provides a promising way to train a machine learning model among mobile devices without collecting their raw data to a central node. During training, proper devices are selected to participate in the training process to avoid model unfairness. In a mobile edge network, participant selection must be considered together with three factors: non-iid datasets possessed by devices, tunable local iterations on devices, and radio resource allocation to counter the impact of time-varying channel conditions on parameter transmissions. Since datasets of devices are given, to ensure model fairness and achieve fast convergence in the FL training process, participants, local iterations, and radio resources must be scheduled jointly in each iteration of FL training. In this paper, the joint scheduling problem is analyzed and formulated. Since it is NP-hard, a heuristic scheduling method called PALORA is designed to conduct joint scheduling of participants, local iterations, and radio resources. PALORA consists of three sequentially interactive function blocks: 1) a pointer network embedded deep reinforcement learning method to select participants, 2) an estimation algorithm to determine the numbers of local iterations, and 3) a breadth-first search method to allocate radio resources to the selected participants. PALORA is evaluated via extensive simulations based on real-world datasets. Results show that it significantly outperforms benchmark approaches.

Efficient urban layout generation is an interesting and important problem in many applications dealing with computer graphics and entertainment. We introduce a novel framework for intuitive and controllable small and large-scale urban layout editing. The key inspiration comes from the observation that cities develop in small incremental changes e.g., a building is replaced, or a new road is created. We introduce a set of atomic operations that consistently modify the city. For example, two buildings are merged, a block is split in two, etc. Our second inspiration comes from volumetric editings, such as clay manipulation, where the manipulated material is preserved. The atomic operations are used in interactive brushes that consistently modify the urban layout. The city is populated with agents. Like volume transfer, the brushes attract or repulse the agents, and blocks can be merged and populated with smaller buildings. We also introduce a large-scale brush that repairs a part of the city by learning style as distributions of orientations and intersections.

Container-based operation system (OS) level virtualization has been adopted by many edge-computing platforms. However, for an edge server, inter-container communications, and container management consume significant CPU resources. Given an application composed of interdependent tasks, the number of such operations is closely related to the dependency between the scheduled tasks. Thus, to improve the execution efficiency of an application in an edge server, task scheduling and task containerizing need to be considered together. To this end, a joint task scheduling and containerizing (JTSC) scheme is developed in this article. Experiments are first carried out to quantify the resource utilization of container operations. System models are then built to capture the features of task execution in containers in an edge server with multiple processors. With these models, joint task scheduling and containerizing is conducted as follows. First, tasks are scheduled without considering containerization, which results in initial schedules. Second, based on system models and guidelines gained from the initial schedules, several containerization algorithms are designed to map tasks to containers. Third, task execution durations are updated by adding the time for inter-container communications, and then the task schedules are updated accordingly. The JTSC scheme is evaluated through extensive simulations. The results show that it reduces inefficient container operations and enhances the execution efficiency of applications by 60 percent.

As an important solution for 3D shape retrieval, a multi-view shape descriptor has achieved impressive performance. One crucial part of view-based shape descriptors is to interpret 3D structures through various 2D observations. Most existing methods like MVCNN believe that a strong classification model trained with deep learning, can often provide an efficient shape embedding for 3D shape retrieval. However, these methods pay much attention to discriminative models and none of them necessarily incorporate the underlying 3D properties of the objects from 2D images. In this paper, we present a novel encoder-decoder recurrent feature aggregation network (ERFA-Net) to address this problem. Aiming at emphasizing the 3D properties of 3D shapes in the fusion of multiple view features, 3D properties prediction tasks are introduced into the 3D shape retrieval. Specifically, an image sequence of the shape is recurrently aggregated into a discriminative shape embedding based on LSTM network, and then this latent shape embedding is trained to predict the original voxel grids and estimate images of unseen viewpoints. This generation task gives an effective supervision which makes the network exploit 3D properties of shapes through various 2D images. Our method achieves the state-of-the-art performance for 3D shape retrieval, on two large-scale 3D shape datasets, ModelNet and ShapeNetCore55. Extensive experiments show that the proposed 3D representation performs robust discrimination against view occlusion, and strong generation ability for various 3D shape tasks.

Multi-view based shape descriptors have achieved impressive performance for 3D shape retrieval. The core of view-based methods is to interpret 3D structures through 2D observations. However, most existing methods pay more attention to discriminative models and none of them necessarily incorporate the 3D properties of the objects. To resolve this problem, we propose an encoder-decoder recurrent feature aggregation network (ERFA-Net) to emphasize the 3D properties of 3D shapes in multi-view features aggregation. In our network, a view sequence of the shape is trained to encode a discriminative shape embedding and estimate unseen rendered views of any viewpoints. This generation task gives an effective supervision which makes the network exploit 3D properties of shapes through various 2D images. During feature aggregation, a discriminative feature representation across multiple views is effectively exploited based on LSTM network. The proposed 3D representation has following advantages against other state-of-the-art: 1) it performs robust discrimination under the existence of noise such as view missing and occlusion, because of the improvement brought by 3D properties. 2) it has strong generative capabilities, which is useful for various 3D shape tasks. We evaluate ERFA-Net on two popular 3D shape datasets, ModelNet and ShapeNetCore55, and ERFA-Net outperforms the state-of-the-art methods significantly. Extensive experiments show the effectiveness and robustness of the proposed 3D representation.