Number Of Rounds Research Articles

Efficient Federated Learning for Feature Aggregation with Heterogenous Edge Devices

Federated learning is a powerful distributed machine learning paradigm for feature aggregation and learning from multiple heterogenous edge devices, due to its ability to keep the privacy of data. However, the training is inefficient for heterogenous devices with considerable communication. Progressive learning is a promising approach for improving the efficiency. Since progressive learning partitions the training process into multiple stages, it is necessary to determine the number of rounds for each stage, and balance the trade-off between saving the energy and improving the model accuracy. Through pilot experiments, we find that the profile which reflects the relationship between round allocation and model quality remains similar in different hyper-parameter configurations, and also observe that the model quality is lossless if the complete model gets sufficient training. Based on the phenomena, we formulate an optimization problem which minimizes the energy consumption of all devices, under the constraint of model quality. We then design a polynomial-time algorithm for the problem. Experimental results demonstrate the superiority of our proposed algorithm under various settings.

Predicting scoring trends in basketball games based on multivariate time series analysis

Abstract In this paper, we first study the evaluation framework of Granger causality between time series. Then, the multivariate time series prediction model is constructed based on transfer entropy and graph attention network. The multivariate time series prediction model architecture has causal branches with transfer entropy, and the prediction module is executed using graph attention network. Finally, the accuracy of basketball game score prediction is analyzed, as well as comparing the accuracy of the multivariate time series prediction model with other models. The number of offensive rounds in basketball games can be predicted with a high accuracy of 67.94% using the multivariate time series prediction model. For the prediction of the scoring trend of player rotation, the ABHJI combination rotating on the court is the most efficient in scoring, and the average value of scoring is as high as 85.463. The average accuracy of the TSP reaches 72.2%, and the model has high feasibility and is suitable for scoring prediction in basketball games.

FedQMIX: Communication-efficient federated learning via multi-agent reinforcement learning

Since the data samples on client devices are usually non-independent and non-identically distributed (non-IID), this will challenge the convergence of federated learning (FL) and reduce communication efficiency. This paper proposes FedQMIX, a node selection algorithm based on multi-agent reinforcement learning(MARL), to address these challenges. Firstly, we observe a connection between model weights and data distribution, and a clustering algorithm can group clients with similar data distribution into the same cluster. Secondly, we propose a QMIX-based mechanism that learns to select devices from clustering results in each communication round to maximize the reward, penalizing the use of more communication rounds and thereby improving the communication efficiency of FL. Finally, experiments show that FedQMIX can reduce the number of communication rounds by 11% and 30% on the MNIST and CIFAR-10 datasets, respectively, compared to the baseline algorithm (Favor).

Deterministic rendezvous in infinite trees

The rendezvous task calls for two mobile agents, starting from different nodes of a network modeled as a graph to meet at the same node. Agents have different labels which are integers from a set {1,…,L}. They wake up at possibly different times and move in synchronous rounds. In each round, an agent can either stay idle or move to an adjacent node. We consider deterministic rendezvous algorithms. The time of such an algorithm is the number of rounds since the wakeup of the earlier agent till the meeting. In most of the literature concerning rendezvous in graphs, the graph is finite and the time of rendezvous depends on its size. This approach is impractical for very large graphs and impossible for infinite graphs. For such graphs it is natural to design rendezvous algorithms whose time depends on the initial distance D between the agents. In this paper we adopt this approach and consider rendezvous in infinite trees. All our algorithms work in finite trees as well. Our main goal is to study the impact of orientation of a tree on the time of rendezvous.We first design a rendezvous algorithm working for unoriented regular trees, whose time is in O(z(D)log⁡L), where z(D) is the size of the ball of radius D, i.e., the number of nodes at distance at most D from a given node. The algorithm works for arbitrary delay between waking times of agents and does not require any initial information about parameters L or D. Its disadvantage is its complexity: z(D) is exponential in D for any degree d>2 of the tree. We prove that this high complexity is inevitable: Ω(z(D)) turns out to be a lower bound on rendezvous time in unoriented regular trees, even for simultaneous start and even when agents know L and D. Then we turn attention to oriented trees. While for arbitrary delay between waking times of agents the lower bound Ω(z(D)) still holds, for simultaneous start the time of rendezvous can be dramatically shortened. We show that if agents know either a polynomial upper bound on L or a linear upper bound on D, then rendezvous can be accomplished in oriented trees in time O(Dlog⁡L), which is optimal. If no such extra knowledge is available, a significant speedup is still possible: in this case we design an algorithm working in time O(D2+log2⁡L).

Off-diagonal online size Ramsey numbers for paths

Consider the following Ramsey game played on the edge set of KN. In every round, Builder selects an edge and Painter colours it red or blue. Builder’s goal is to force Painter to create a red copy of a path Pk on k vertices or a blue copy of Pn as soon as possible. The online (size) Ramsey number r̃(Pk,Pn) is the number of rounds in the game provided Builder and Painter play optimally. We prove that r̃(Pk,Pn)≤(5/3+o(1))n provided k=o(n) and n→∞. We also show that r̃(P4,Pn)≤⌈7n/5⌉−1 for n≥10, which improves the upper bound obtained by J. Cyman, T. Dzido, J. Lapinskas, and A. Lo and implies their conjecture that r̃(P4,Pn)=⌈7n/5⌉−1.

Mean Field Multi-Agent Reinforcement Learning Method for Area Traffic Signal Control

Reinforcement learning is an effective method for adaptive traffic signal control in urban transportation networks. As the number of training rounds increases, the optimal control strategy is learned, and the learning capabilities of deep neural networks are further enhanced, thereby avoiding the limitations of traditional signal control methods. However, when faced with the sequential decision tasks of regional signal control, it encounters issues such as the curse of dimensionality and environmental non-stationarity. To address the limitations of traditional reinforcement learning algorithms applied to multiple intersections, the mean field theory is applied. This models the traffic signal control problem at multiple intersections within a region as interactions between individual intersections and the average effects of neighboring intersections. By decomposing the Q-function through bilateral estimation between the agent and its neighbors, this method reduces the complexity of interactions between agents while preserving global interactions between the agents. A traffic signal control model based on Mean Field Multi-Agent Reinforcement Learning (MFMARL) was constructed, containing two algorithms: Mean Field Q-Network Area Traffic Signal Control (MFQ-ATSC) and Mean Field Actor-Critic Network Area Traffic Signal Control (MFAC-ATSC). The model was validated using the SUMO simulation platform. The experimental results indicate that across different metrics, such as average speed, the mean field reinforcement learning method outperforms classical signal control methods and several existing approaches.

PRICE CLUSTERING BEHAVIOR IN VIRTUAL REAL ESTATE MARKETS

We analyze 21,209 intraday transactions in the virtual real estate market and document significant price clustering at round numbers 0, 00, and 000 as ending digits, consistent with the negotiation hypothesis. The clustering increases with price level and pricing uncertainty proxied by the number of buyers and sellers in the NFT market. Moreover, market venue influences price clustering dynamics. Digits 9, 99, and 999 as ending prices are overrepresented in the sample, consistent with the left digit effects. However, we do not find support for the psychologically feeling right hypothesis or the strategic trading hypothesis.

An Optimization Method for Non-IID Federated Learning Based on Deep Reinforcement Learning.

Federated learning (FL) is a distributed machine learning paradigm that enables a large number of clients to collaboratively train models without sharing data. However, when the private dataset between clients is not independent and identically distributed (non-IID), the local training objective is inconsistent with the global training objective, which possibly causes the convergence speed of FL to slow down, or even not converge. In this paper, we design a novel FL framework based on deep reinforcement learning (DRL), named FedRLCS. In FedRLCS, we primarily improved the greedy strategy and action space of the double DQN (DDQN) algorithm, enabling the server to select the optimal subset of clients from a non-IID dataset to participate in training, thereby accelerating model convergence and reaching the target accuracy in fewer communication epochs. In simulation experiments, we partition multiple datasets with different strategies to simulate non-IID on local clients. We adopt four models (LeNet-5, MobileNetV2, ResNet-18, ResNet-34) on the four datasets (CIFAR-10, CIFAR-100, NICO, Tiny ImageNet), respectively, and conduct comparative experiments with five state-of-the-art non-IID FL methods. Experimental results show that FedRLCS reduces the number of communication rounds required by 10-70% with the same target accuracy without increasing the computation and storage costs for all clients.

Language and private debt renegotiation

AbstractWe study how language affects private debt renegotiation. We predict that stronger future time reference (FTR) languages alter the importance of renegotiation risk by lowering the perceived value of loan renegotiation. We test this hypothesis on a sample of 6500 loans issued to European firms between 1999 and 2017. We find that the use of a stronger FTR language decreases the likelihood of renegotiation and the number of renegotiation rounds. These findings are robust to several FTR proxies, various specifications including loan, borrower, and country‐level variables, and potential mitigation effects from specific loan, country, or time effects. They suggest that linguistic structure influences the renegotiation process of private debt contracts.

Numerical investigation of battery thermal management by using helical fin and composite phase change material

Li-ion batteries have become an indispensable part of modern life thanks to their portability, energy density, fast charge-discharge capability and various application areas. The thermal performance of the PCM promising method for battery thermal management systems (BTMS) is limited due to weak thermal conductivity property. The aim of this study is to investigate the contribution to the performance of BTMS by using innovative helical fins to improve the heat transfer performance of PCM due to its low thermal conductivity. Besides, the study includes composite phase change material (CPCM) due to advantages in practice such as form stability, mechanical behavior and higher thermal conductivity compared to PCM. In this scope, a hybrid system including helical fin and CPCM, paraffin-based expanded graphite (EG), is numerically investigated to ensure thermal control of a Li-ion battery in this study. Results shows that safe operating time is prolonged by increasing the widths and the number of rounds of the helical fin. However, the contribution of the EG fractions on the BTMS is ordered as an EG fraction of 12 % > 6 % > 20 % > 0 % due to a balance between the thermal conductivity and the latent heat property. The longest safe operating time is obtained by using Fin_W6-NR3 for EG fraction of 12 % in the CPCM as 9316 s corresponding to 244.0 % higher than the case of EG = 0 % and no fin used.

Leveraging Larger AES Keys in LoRaWAN: A Practical Evaluation of Energy and Time Costs.

Internet of Things (IoT) devices increasingly contribute to critical infrastructures, necessitating robust security measures. LoRaWAN, a low-power IoT network, employs the Advanced Encryption Standard (AES) with a 128-bit key for encryption and integrity, balancing efficiency and security. As computational capabilities of devices advance and recommendations for stronger encryption, such as AES-256, emerge, the implications of using longer AES keys (192 and 256 bits) on LoRaWAN devices' energy consumption and processing time become crucial. Despite the significance of the topic, there is a lack of research on the implications of using larger AES keys in real-world LoRaWAN settings. To address this gap, we perform extensive tests in a real-world LoRaWAN environment, modifying the source code of both a LoRaWAN end device and open-source server stack to incorporate larger AES keys. Our results show that, while larger AES keys increase both energy consumption and processing time, these increments are minimal compared to the time on air. Specifically, for the maximum payload size we used, when comparing AES-256 to AES-128, the additional computational time and energy are, respectively, 750 ms and 236 μJ. However, in terms of time on air costs, these increases represent just 0.2% and 0.13%, respectively. Our observations confirm our intuition that the increased costs correlate to the number of rounds of AES computation. Moreover, we formulate a mathematical model to predict the impact of longer AES keys on processing time, which further supports our empirical findings. These results suggest that implementing longer AES keys in LoRaWAN is a practical solution enhancing its security strength while not significantly impacting energy consumption or processing time.

FPCNN: A fast privacy-preserving outsourced convolutional neural network with low-bandwidth

Convolutional neural networks (CNNs) have been widely employed in image and video recognition tasks due to their superior performance. With the increasing popularity of machine learning as a service, CNN model owners often outsource their models to the cloud to provide inference services. However, the user data and outsourced models processed by cloud-based CNN inference applications suffer from privacy leakage issues. Therefore, several studies have proposed different privacy-preserving CNN architectures. Most of these approaches are based on heavyweight cryptographic methods, which require significant computational resources. In this study, we propose a novel scheme, a Fast Privacy-Preserving Outsourced Convolutional Neural Network with Low Bandwidth (FPCNN), to provide secure inference services on the cloud using outsourced data and models. Specifically, for linear CNN layers, FPCNN significantly reduces the bandwidth requirements between servers by applying scaled noise models and scaled secret sharing. For nonlinear CNN layers, we first optimize the multiplication operation under secret sharing to save bandwidth. Subsequently, based on the nonlinear input characteristics of CNNs, we introduce a secure parallel rectified linear unit (ReLU) computation protocol that significantly reduces the number of communication rounds. Finally, the experimental results demonstrate that FPCNN achieves a speedup of 2.66× and a 32.8× reduction in bandwidth compared with state-of-the-art methods.

Image-based crop disease detection with federated learning

Crop disease detection and management is critical to improving productivity, reducing costs, and promoting environmentally friendly crop treatment methods. Modern technologies, such as data mining and machine learning algorithms, have been used to develop automated crop disease detection systems. However, centralized approach to data collection and model training induces challenges in terms of data privacy, availability, and transfer costs. To address these challenges, federated learning appears to be a promising solution. In this paper, we explored the application of federated learning for crop disease classification using image analysis. We developed and studied convolutional neural network (CNN) models and those based on attention mechanisms, in this case vision transformers (ViT), using federated learning, leveraging an open access image dataset from the “PlantVillage” platform. Experiments conducted concluded that the performance of models trained by federated learning is influenced by the number of learners involved, the number of communication rounds, the number of local iterations and the quality of the data. With the objective of highlighting the potential of federated learning in crop disease classification, among the CNN models tested, ResNet50 performed better in several experiments than the other models, and proved to be an optimal choice, but also the most suitable for a federated learning scenario. The ViT_B16 and ViT_B32 Vision Transformers require more computational time, making them less suitable in a federated learning scenario, where computational time and communication costs are key parameters. The paper provides a state-of-the-art analysis, presents our methodology and experimental results, and concludes with ideas and future directions for our research on using federated learning in the context of crop disease classification.

Conflict-based negotiation strategy for human-agent negotiation

Day by day, human-agent negotiation becomes more and more vital to reach a socially beneficial agreement when stakeholders need to make a joint decision together. Developing agents who understand not only human preferences but also attitudes is a significant prerequisite for this kind of interaction. Studies on opponent modeling are predominantly based on automated negotiation and may yield good predictions after exchanging hundreds of offers. However, this is not the case in human-agent negotiation in which the total number of rounds does not usually exceed tens. For this reason, an opponent model technique is needed to extract the maximum information gained with limited interaction. This study presents a conflict-based opponent modeling technique and compares its prediction performance with the well-known approaches in human-agent and automated negotiation experimental settings. According to the results of human-agent studies, the proposed model outpr erforms them despite the diversity of participants’ negotiation behaviors. Besides, the conflict-based opponent model estimates the entire bid space much more successfully than its competitors in automated negotiation sessions when a small portion of the outcome space was explored. This study may contribute to developing agents that can perceive their human counterparts’ preferences and behaviors more accurately, acting cooperatively and reaching an admissible settlement for joint interests.

The energy complexity of diameter and minimum cut computation in bounded-genus networks

This paper investigates the energy complexity of distributed graph problems in multi-hop radio networks, where the energy cost of an algorithm is measured by the maximum number of awake rounds of a vertex. Recent works revealed that some problems, such as broadcast, breadth-first search, and maximal matching, can be solved with energy-efficient algorithms that consume only polylog⁡n energy. However, there exist some problems, such as computing the diameter of the graph, that require Ω(n) energy to solve. To improve energy efficiency for these problems, we focus on a special graph class: bounded-genus graphs. We present algorithms for computing the exact diameter, the exact global minimum cut size, and a (1±ϵ)-approximate s-t minimum cut size with O˜(n) energy for bounded-genus graphs. Our approach is based on a generic framework that divides the vertex set into high-degree and low-degree parts and leverages the structural properties of bounded-genus graphs to control the number of certain connected components in the subgraph induced by the low-degree part.

Optimization of the Drawing Process of Low–Middle-Thickness Top Coal Based on a Simulation

Based on a theoretical analysis and simulation, the automated drawing characteristics for low–middle-thickness top coal are investigated, and the drawing process of top coal is optimized. For top coal with a 3 m thickness, a comparison is made between traditional and automated drawing process of top coal, which is used to highlight the superiority of automated drawing process. Then, for the automated top coal drawing, the effects of the round number on the drawing results of the entire working face are investigated, and the advantages and disadvantages of the multi-round drawing process are compared. Finally, the coal drawing results combined with automated and traditional methods are discussed, and an ideal optimized coal drawing process is proposed. The research results indicate that, for top coal with a 3 m thickness, the recovery rate of top coal by automated one-round drawing is slightly higher than that of traditional drawing, the drawing amount and drawing time of various drawing openings are relatively uniform for automated drawing, while the traditional coal drawing process shows fluctuating characteristics. For automated multi-round drawing, the top coal recovery rate decreases with an increase in the round number; however, the rock mixed rate in the drawn top coal is lower, and the drawing results at various drawing openings become more uniform. The combined automated–traditional drawing can achieve layered drawing of the top coal, avoid gangue drawing, and maintain a relatively high recovery rate, which is the optimal drawing process for this research project.

Deterministic treasure hunt and rendezvous in arbitrary connected graphs

Treasure hunt and rendezvous are fundamental tasks performed by mobile agents in graphs. In treasure hunt, an agent has to find an inert target (called treasure) situated at an unknown node of the graph. In rendezvous, two agents, initially located at distinct nodes of the graph, traverse its edges in synchronous rounds and have to meet at some node. We assume that the graph is connected (otherwise none of these tasks is feasible) and consider deterministic treasure hunt and rendezvous algorithms. The time of a treasure hunt algorithm is the worst-case number of edge traversals performed by the agent until the treasure is found. The time of a rendezvous algorithm is the worst-case number of rounds since the wakeup of the earlier agent until the meeting.To the best of our knowledge, all known treasure hunt and rendezvous algorithms rely on the assumption that degrees of all nodes are finite, even when the graph itself may be infinite. In the present paper we remove this assumption for the first time, and consider both above tasks in arbitrary connected graphs whose nodes can have either finite or countably infinite degrees. Our main result is the first universal treasure hunt algorithm working for arbitrary connected graphs. We prove that the time of this algorithm has optimal order of magnitude among all possible treasure hunt algorithms working for arbitrary connected graphs. As a consequence of this result we obtain the first universal rendezvous algorithm working for arbitrary connected graphs. The time of this algorithm is polynomial in a lower bound holding in many graphs, in particular in the tree all of whose degrees are infinite.

Multipass wide-field phase imager.

Advances in optical imaging always look for an increase in sensitivity and resolution among other practicability aspects. Within the same scope, in this work we report a versatile interference contrast imaging technique, with high phase sensitivity and a large field-of-view of several mm2. Sensitivity is increased through the use of a self-imaging non-resonant cavity, which causes photons to probe the sample in multiple rounds before being detected, where the configuration can be transmissive or reflective. Phase profiles can be resolved individually for each round thanks to a specially designed single-photon camera with time-of-flight capabilities and true pixels-off gating. Measurement noise is reduced by novel data processing combining the retrieved sample profiles from multiple rounds. Our protocol is especially useful under extremely low light conditions as required by biological or photo-sensitive samples. Results demonstrate more than a four-fold reduction in phase measurement noise, compared to single round imaging, and values close to the predicted sensitivity in case of the best possible cavity configuration, where all photons are maintained until n rounds. We also find good agreement with the theoretical predictions for low number of rounds, where experimental imperfections would play a minor role. The absence of a laser or cavity lock-in mechanism makes the technique an easy to use inspection tool.

Hierarchical energy-saving routing algorithm using fuzzy logic in wireless sensor networks

Currently, sensor energy assembly in wireless sensor networks is limited, and clustering methods are not effective to improve sensor energy consumption rate. Thus, a hierarchical energy-saving routing algorithm based on fuzzy logic was constructed by considering three aspects: residual energy value, centrality, and distance value between nodes and base stations. The remaining sensor nodes selected by fuzzy logic algorithm have a longer time to live and greater residual energy than those selected by low-power adaptive clustering hierarchical protocol algorithm, fuzzy unequal clustering algorithm, and fuzzy logic cluster head election algorithm. For network life cycle, the number of rounds in which the first dead node appears, in descending order, is studied: energy-saving routing algorithm (400 rounds) > new geographic cellular structure algorithm (300 rounds) > virtual grid based dynamic routes adjustment algorithm (100 rounds). Under the same experimental round, energy-saving routing algorithm’s remaining energy curve always reaches its maximum. The energy-saving routing algorithm by fuzzy logic constructed by this research institute can significantly improve network energy utilization, which has certain reference value.

LEACH-D: A low-energy, low-delay data transmission method for industrial internet of things wireless sensors

In recent years, the Internet of Things (IoT) has experienced extensive adoption in industrial environments, healthcare, smart cities, and more, playing a vital role in these domains. Within IoT-based systems, wireless sensor networks (WSNs) have emerged as a crucial method for collecting peripheral environmental data within industries, owing to their self-organizational attributes. Nevertheless, the enormous volume of heterogeneous data from various sensing devices presents many challenges for IoT-enabled WSNs, encompassing high transmission delay times (TD) and excessive battery energy consumption (EC). To address these challenges, it is imperative to prioritize efficiency and optimize energy utilization. Moreover, enhancing energy efficiency within the Industrial Internet of Things (IIoT) realm hinges significantly on factors such as data transmission modes and the allocation of cluster head nodes. Numerous researchers have proposed algorithms to minimize transmission time and energy consumption, specifically focusing on industrial environments. This paper introduces an inventive clustering-based data transmission algorithm for IIoT, LEACH-D, to enhance efficiency. The LEACH-D algorithm improves the transmission task duration while maintaining consistent battery energy consumption. It also seeks to elevate performance in metrics such as average transmission time during the first node death (FND). Numerous experimental results provide strong evidence that the algorithm introduced in this paper has effectively reduced the average transmission time by remarkable percentages: 51.32%, 12.12%, 12.96%, and 5.42%, while simultaneously increasing the number of FND rounds by significant margins: 222.43%, 36.63%, 33.72%, and 7.81%, respectively. These improvements stand in stark contrast to the performance of existing algorithms, including FREE_MODE, LEACH, EE-LEACH, and ETH-LEACH.