Knowledge Distillation (KD) is considered as an efficient way to replace the parameter averaging in federated learning, aiming to handle the clients with heterogeneous model architectures. Relying on the prepared distillation datasets across clients and the server, KD may encounter impractical difficulties in real-world implementations. Existing works explore the data-free KD in federated learning, which generates the distillation datasets on-site. However, the distillation datasets with global data distribution generated by these state-of-the-art schemes cannot be adapted to local non-IID data. In this article, we propose a new Personalized Data-Free Knowledge Distillation, namely PDKD, for federated learning under heterogeneous models and data. PDKD solves the problem of model drift caused by the inconsistent distribution of distillation datasets and the local data by generating personalized distillation datasets for each client while protecting client data privacy. In addition, we design a distillation dataset update scheme that maximizes the difference between teacher and client outputs on distillation datasets to accomplish deeper knowledge transfer. Furthermore, in order to accomplish the co-evolution of the teacher model and the clients’ model, PDKD incorporates a mutual distillation scheme. Numerous experiments show that PDKD significantly outperforms several state-of-the-art algorithms, with an 18% improvement in prediction accuracy and has a much lower communication cost than the compared algorithms.

Federated learning enables multiple clients to collaboratively learn machine learning models in a privacy-preserving manner. However, in real-world scenarios, a key challenge encountered in federated learning is the statistical heterogeneity among clients. Existing work mainly focused on a single global model shared across the clients, making it hard to generalize well to all clients due to the large discrepancy in the data distributions. To address this challenge, we propose pFedLT , a novel approach that can adapt the single global model to different data distributions. Specifically, we propose to perform a pluggable layer-wise transformation during the local update phase based on scaling and shifting operations. In particular, these operations are learned with a meta-learning strategy. By doing so, pFedLT can capture the diversity of data distribution among clients, therefore, can generalize well even when the data distributions among clients exhibit high statistical heterogeneity. We conduct extensive experiments on synthetic and real-world datasets (MNIST, Fashion_MNIST, CIFAR-10, and Office+Caltech10) under different Non-IID settings. Experimental results demonstrate that pFedLT significantly improves the model accuracy by up to 11.67% and reduces the communication costs compared with state-of-the-art approaches.

Popular ride-hailing products, such as DiDi, Uber and Lyft, provide people with transportation convenience. Pricing, order dispatching and vehicle repositioning are three tasks with tight correlation and complex interactions in ride-hailing platforms, significantly impacting each other’s decisions and demand distribution or supply distribution. However, no past work considered combining the three tasks to improve platform efficiency. In this paper, we exploit to optimize pricing, dispatching and repositioning strategies simultaneously. Such a new multi-stage decision-making problem is quite challenging because it involves complex coordination and lacks a unified problem model. To address this problem, we propose a novelJoint optimization framework ofPricing,Dispatching andRepositioning (JPDR) integrating contextual bandit and multi-agent deep reinforcement learning. JPDR consists of two components, including a Soft Actor-Critic (SAC)-based centralized policy for dispatching and repositioning and a pricing strategy learned by a multi-armed contextual bandit algorithm based on the feedback from the former. The two components learn in a mutually guided way to achieve joint optimization because their updates are highly interdependent. Based on real-world data, we implement a realistic environment simulator. Extensive experiments conducted on it show our method outperforms state-of-the-art baselines in terms of both gross merchandise volume and success rate.

Personalized federated learning (PFL) has emerged as a paradigm to provide a personalized model that can fit the local data distribution of each client. One natural choice for PFL is to leverage the fast adaptation capability of meta-learning, where it first obtains a single global model, and each client achieves a personalized model by fine-tuning the global one with its local data. However, existing meta-learning-based approaches implicitly assume that the data distribution among different clients is similar, which may not be applicable due to the property of data heterogeneity in federated learning. In this work, we propose a Group-based Federated Meta-Learning framework, called G-FML , which adaptively divides the clients into groups based on the similarity of their data distribution, and the personalized models are obtained with meta-learning within each group. In particular, we develop a simple yet effective grouping mechanism to adaptively partition the clients into multiple groups. Our mechanism ensures that each group is formed by the clients with similar data distribution such that the group-wise meta-model can achieve “personalization” at large. By doing so, our framework can be generalized to a highly heterogeneous environment. We evaluate the effectiveness of our proposed G-FML framework on three heterogeneous benchmarking datasets. The experimental results show that our framework improves the model accuracy by up to 13.15% relative to the state-of-the-art federated meta-learning.

Autonomous on-demand services, such as GOGOX (formerly GoGoVan) in Hong Kong, provide a platform for users to request services and for suppliers to meet such demands. In such a platform, the suppliers have autonomy to accept or reject the demands to be dispatched to him/her, so it is challenging to make an online matching between demands and suppliers. Existing methods use round-based approaches to dispatch demands. In these works, the dispatching decision is based on the predicted response patterns of suppliers to demands in the current round, but they all fail to consider the impact of future demands and suppliers on the current dispatching decision. This could lead to taking a suboptimal dispatching decision from the future perspective. To solve this problem, we propose a novel demand dispatching model using deep reinforcement learning. In this model, we make each demand as an agent. The action of each agent, i.e., the dispatching decision of each demand, is determined by a centralized algorithm in a coordinated way. The model works in the following two steps. (1) It learns the demand’s expected value in each spatiotemporal state using historical transition data. (2) Based on the learned values, it conducts a Many-To-Many dispatching using a combinatorial optimization algorithm by considering both immediate rewards and expected values of demands in the next round. In order to get a higher total reward, the demands with a high expected value (short response time) in the future may be delayed to the next round. On the contrary, the demands with a low expected value (long response time) in the future would be dispatched immediately. Through extensive experiments using real-world datasets, we show that the proposed model outperforms the existing models in terms of Cancellation Rate and Average Response Time.