Federated learning (FL) is a privacy-motivated paradigm for distributed training of deep learning models, as it allows a large number of clients to collaboratively train a shared global model without centralizing their private data. Since its debut with Federated Averaging as the first server aggregation algorithm, many FL algorithms have been proposed to improve performance. Yet, these algorithms are rarely benchmarked and compared under the same open-source framework and controlled configurations, and their performance claims can be difficult to substantiate in fair and reproducible studies. In this paper, we evaluate a curated and representative collection of FL algorithms in the same open-source benchmarking framework so that they can be compared fairly at scale in a reproducible fashion. To achieve this objective, we presentPlato, an open-source FL research framework that we have designed and implemented from scratch. WithPlato, we evaluate and compare algorithms spanning (1) server aggregation; (2) client training customization; (3) client selection, in both synchronous and asynchronous settings; (4) personalized federated learning; and (5) communication efficiency and payload processing. Across diverse experimental scenarios (tasks, client populations, and data distributions), we report findings and practical insights, including pitfalls and confounding factors that can lead to misleading conclusions if not reported carefully. Under our unified experimental settings and time model, Federated Averaging with random client selection remains a strong baseline and is often competitive with more complex alternatives.

Federated learning is a decentralized framework that enables client participation in collaborative learning without centralized data collection. However, the framework is susceptible to suboptimal model convergence induced by heterogeneity among the client datasets. These discrepancies, including label imbalance, dissimilarity in data distributions, and uneven data volumes between clients, may cause disagreements among local client updates, affecting the ability of the global model to converge effectively during aggregation. We suggest that one potential solution to this problem lies in weighting the model aggregation by client importance and client-to-client relationships. Based on this idea, we propose HyperGenFL (HG-FL), a hypernetwork that generates aggregation weights from learnable client embeddings without requiring any training or benchmarking data. HG-FL utilizes the attention mechanism to capture inter-client relationships based on learnable client-specific embeddings in order to generate model aggregation weights dynamically during federated learning. By guiding the aggregation process with these learnable relationships between local models, HG-FL reduces update conflicts and improves global model performance. We assess HG-FL under various data-heterogeneous environments based on different benchmark datasets including Fashion-MNIST, CIFAR10, CIFAR100 and Tiny-ImageNet. Experimental results demonstrate that HG-FL can achieve superior performance over a range of existing baseline methods under challenging cases with various heterogeneous environments, large models and a large number of clients.

Directed Acyclic Graph (DAG)-based blockchain (a.k.a distributed ledger) has become prevalent for supporting highly concurrent applications. Its inherent parallel data structure accelerates block generation significantly, shifting the bottleneck from performance to storage scalability. An intuitive solution is to apply state sharding that divides the entire ledger (i.e., transactions and states) into multiple shards. While each node only stores proportional transactions, it suffers from the challenges of storing and ensuring the processing consistency of cross-shard transactions. In this paper, we propose SharDAG, a new mechanism that leverages adaptive sharding for DAG-based blockchains to achieve high performance and strong consistency. The key idea of SharDAG is to exploit unique characteristics - silent assets - and design a lightweight processing mechanism based on avatar account caching. Furthermore, we design a Byzantine resilient cross-shard verification mechanism with a theoretically optimal number of participating nodes, which guarantees the consistency and security of avatar account aggregation. Our comprehensive evaluations on real-world workloads demonstrate that SharDAG presents up to 3.8 x throughput improvement compared to the state-of-the-art and reduces the storage overhead of cross-shard transactions.

Federated learning is a decentralized approach to training machine learning models while preserving data privacy. To accommodate data heterogeneity among clients, a longstanding issue in Federated Learning, many Personalized Federated Learning (PFL) strategies decompose each client model into global modules, which are collaboratively learned by all clients and the server, and local modules, which are only trained locally on private data. While these strategies require every client to participate in training, in reality, many client devices lack sufficient data or computing resources to perform meaningful local training, making it difficult to achieve personalization for every client. In this paper, we present HyperFLoRA, a PFL framework that leverages knowledge learned from training-capable clients to enable the immediate creation of personalized models for training-incapable or new clients. HyperFLoRA uses adapters for personalization to minimize communication costs and client training workload while employing a trainable hypernetwork to generate personalized adapter weights for each client using minimal client statistical information. From experiments conducted on both convolutional and Transformer neural networks, HyperFLoRA can achieve superior model personalization performance for new clients that did not participate in training than conventional PFL methods, while significantly reducing training-related communication costs and client workload.

Distributed applications that utilizeheterogeneousblockchain systems have the potential to be widely deployed. In such applications, users from different blockchains can transact with one another throughcross-chain transactions. There are two essential features of particular relevance for those applications during cross-chain transactions: theatomicityin that either all or none of the blockchains involved confirm a cross-chain transaction, theconfidentialityin that a blockchain involved in a cross-chain transaction is only accessible for designated users. Existing cross-chain proposals have largely relied on permissioned blockchains to ensure confidentiality. However, we found that failures could occur when reading or writing information during transaction confirmations across permissioned blockchains, namely read/write (r/w) failures, which can lead to the violation of atomicity. In this paper, we propose a novel mechanism,Unity, to ensure both atomicity and confidentiality of cross-chain transactions under r/w failures by leveraging permissioned blockchains. When failures occur in reading or writing data,Unityclassifies the data into two categories based on its status - whether data is the latest version or not, and presents different solutions for atomicity. Specifically, when data is not the latest, we design a four-phase-commit protocol4pc, in which consensus on confirming or aborting a cross-chain transaction can be achieved. If data is the latest when r/w failures occur, we propose a smart contract based solution SSC. We examine the effectiveness ofUnitytheoretically and through experiments. With a failure probability of 0.7,Unityachieves$98\%$more atomic cross-chain transactions when compared with the state-of-the-art cross-chain platform, Hyperservice.

Heterogeneous blockchains are expected to be increasingly deployed in real-world applications, making cross-chain transaction confirmations essential. Currently, confirmations for cross-chain transactions are usually accomplished through an intermediary, such as a relay chain, which may well become a performance bottleneck. Sharding has been widely used to improve the blockchain throughput through parallel transaction validations by distributing transactions into multiple sub-communities, Yet, when sharding technique is directly applied over a relay chain, it results in an excessive number of cross-shard transactions, offsetting the throughput improvement.In this paper, we propose Sliver, a novel transaction distribution mechanism specifically designed for improving the relay chain throughput for the first time. We first capture and leverage the unique characterization of transaction dependency on the relay chain, and place those transactions with dependency into one shard. Consequently, this completely eliminates cross-shard transactions. However, due to the varying nature of transaction dependency, such a transaction placement can lead to a highly skewed distribution in terms of the number of transactions (i.e., shard size) to be validated in different shards, which negatively affects the relay chain throughput. We proceed to formulate the transaction distribution as an integer optimization problem with a lexicographical minimization objective for achieving a balanced shard size. While such a problem is proved to be NP-hard, we are able to mathematically transform it to a linear programming (LP) formulation by incorporating several unique properties in the integer optimization formulation, which can then be efficiently solved using off-the-shelf LP solvers. Theoretical and experimental analyses show that Sliver is extremely efficient in solving the assignment problem and the throughput can be 5 × that of the state-of-the-art under various configurations.

In most blockchain-based application scenarios, a complete application logic consists of multiple continuous transactions, in which the initiation of one transaction depends on the confirmation result of the previous one. This mandates that continuous transactions must be processed in the correct order. Unfortunately, existing chain-based blockchains fail to effectively support continuous transaction processing due to considerable latency in confirming continuous transactions. Recent studies shifted from chain-based blockchains toDirected Acyclic Graph(DAG) based blockchains, which reduced transaction confirmation latencies. However, DAG-based blockchains store transactions in an out-of-order manner that leads to unordered transaction processing. To address this challenge, we propose FLUID, a new DAG-based blockchain that supports continuous transaction processing while delivering high performance. The fundamental idea of FLUID is to design a transaction dependency tracking structure to ensure that continuous transactions can be processed in the correct order. FLUID utilizes a conflict resolution mechanism to provide instant confirmation and to support concurrent transaction processing with lower latencies. In addition, FLUID builds a checkpoint-based verification mechanism to achieve deterministic consensus on transaction processing results in the DAG. Extensive experiments demonstrate that our proposed FLUID can improve the throughput over state-of-the-art OHIE by 66% with two orders of magnitude lower latencies.

With their instant transaction confirmation and high scalability, payment channel networks (PCNs), running off-chain and in parallel with blockchain systems, have recently attracted a substantial amount of research attention. It has been shown that there exists a significant gap between the theoretically optimal performance and the performance achievable given the stringent privacy requirements in practice. However, it remains unclear what the fundamental performance limits and key factors involved are, which turns out to be a challenging problem due to the unique characteristics in PCNs. In this paper, we, for the first time, develop a mathematical model capturing the PCN performance, and examine the impact from a number of factors including channel capacity and transactions. We are articularly interested in obtaining the gap between the theoretically optimal performance and the performance achievable in practice, which characterizes the design space in PCNs for scheduling transactions. Specifically, we derive how different transactions and channel capacities affect the PCN performance and the performance gap. Our analytical characterization of PCNs offers an in-depth understanding on their fundamental trade-off, and provides important insights on the design of PCNs.

Influence maximization in social networks is of great importance for marketing new products. Signed social networks with both positive (friends) and negative (foes) relationships pose new challenges and opportunities, since the influence of negative relationships can be leveraged to promote information propagation. In this paper, we study the problem of influence maximization for advertisement recommendation in signed social networks. We propose a new framework to characterize the information propagation process in signed social networks, which models the dynamics of individuals' beliefs and attitudes towards the advertisement based on recommendations from both positive and negative neighbours. To achieve influence maximization in signed social networks, we design a novel Signed-PageRank (SPR) algorithm, which selects the initial seed nodes by jointly considering their positive and negative connections with the rest of the network. Our extensive experimental results confirm that our proposed SPR algorithm can effectively and efficiently influence a broader range of individuals in the signed social networks than benchmark algorithms on both synthetic and real datasets.

Current blockchain systems suffer from a number of inherent drawbacks in its scalability, latency, and processing throughput. By enabling parallel confirmations of transactions, sharding has been proposed to mitigate these drawbacks, which usually requires frequent communication among miners through a separate consensus protocol.In this paper, we propose, analyze, and implement a new distributed and dynamic sharding system to substantially improve the throughput of blockchain systems based on smart contracts, while requiring minimum cross-shard communication. Our key observation is that transactions sent by users who only participate in a single smart contract can be validated and confirmed independently without causing double spending. Therefore, the natural formation of a shard is to surround one smart contract to start with. The complication lies in the different sizes of shards being formed, in which a small shard with few transactions tends to generate a large number of empty blocks resulting in a waste of mining power, while a large shard adversely affects parallel confirmations. To overcome this problem, we propose an inter-shard merging algorithm with incentives to encourage small shards to merge with one another and form a larger shard, an intra-shard transaction selection mechanism to encourage miners to select different subsets of transactions for validation, as well as a parameter unification method to further improve these two algorithms to reduce the communication cost and improve system reliability.We analyze our proposed algorithms using the game theoretic approach, and prove that they converge to a Nash Equilibrium. We also present a security analysis on our sharding design, and prove that it resists adversaries who occupy at most 33% of the computation power. We have implemented our designs on go-Ethereum 1.8.0 and evaluated their performance using both real-world blockchain transactions and large-scale simulations. Our results show that throughput has been improved by 7.2×, and the number of empty blocks has been reduced by 90%.