Fair Trade-off Research Articles

Dual temporal memory network with high-order spatio-temporal graph learning for video object segmentation

Typically, Video Object Segmentation (VOS) always has the semi-supervised setting in the testing phase. The VOS aims to track and segment one or several target objects in the following frames in the sequence, only given the ground-truth segmentation mask in the initial frame. A fundamental issue in VOS is how to best utilize the temporal information to improve the accuracy. To address the aforementioned issue, we provide an end-to-end framework that simultaneously extracts long-term and short-term historical sequential information to current frame as temporal memories. The integrated temporal architecture consists of a short-term and a long-term memory modules. Specifically, the short-term memory module leverages a high-order graph-based learning framework to simulate the fine-grained spatial–temporal interactions between local regions across neighboring frames in a video, thereby maintaining the spatio-temporal visual consistency on local regions. Meanwhile, to relieve the occlusion and drift issues, the long-term memory module employs a Simplified Gated Recurrent Unit (S-GRU) to model the long evolutions in a video. Furthermore, we design a novel direction-aware attention module to complementarily augment the object representation for more robust segmentation. Our experiments on three mainstream VOS benchmarks, containing DAVIS 2017, DAVIS 2016, and Youtube-VOS, demonstrate that our proposed solution provides a fair tradeoff performance between both speed and accuracy.

Efficiency Versus Fairness Tradeoffs in Algorithm-Based Personnel Selection

Efficiency Versus Fairness Tradeoffs in Algorithm-Based Personnel Selection

Investigating and Mitigating the Performance–Fairness Tradeoff via Protected-Category Sampling

Machine learning algorithms have become common in everyday decision making, and decision-assistance systems are ubiquitous in our everyday lives. Hence, research on the prevention and mitigation of potential bias and unfairness of the predictions made by these algorithms has been increasing in recent years. Most research on fairness and bias mitigation in machine learning often treats each protected variable separately, but in reality, it is possible for one person to belong to multiple protected categories. Hence, in this work, combining a set of protected variables and generating new columns that separate these protected variables into many subcategories was examined. These new subcategories tend to be extremely imbalanced, so bias mitigation was approached as an imbalanced classification problem. Specifically, four new custom sampling methods were developed and investigated to sample these new subcategories. These new sampling methods are referred to as protected-category oversampling, protected-category proportional sampling, protected-category Synthetic Minority Oversampling Technique (PC-SMOTE), and protected-category Adaptive Synthetic Sampling (PC-ADASYN). These sampling methods modify the existing sampling method by focusing their sampling on the new subcategories rather than the class label. The impact of these sampling strategies was then evaluated based on classical performance and fairness in classification settings. Classification performance was measured using accuracy and F1 based on training univariate decision trees, and fairness was measured using equalized odd differences and statistical parity. To evaluate the impact of fairness versus performance, these measures were evaluated against decision tree depth. The results show that the proposed methods were able to determine optimal points, whereby fairness was increased without decreasing performance, thus mitigating any potential performance–fairness tradeoff.

Power allocation scheme for sum rate and fairness trade-off in downlink NOMA networks

Non-orthogonal multiple access (NOMA) is an essential enabler technology that is expected to help satisfy the key requirements of increased system throughput in future wireless networks. However, another equally important aspect that should go hand-in-hand with system throughput is user fairness for any network. But, to the best of our knowledge, even though there have been works that look at system throughput or user fairness maximization for NOMA-based networks, they looked at these as a single objective optimization problem, where one is the objective and the other is one of the constraints. However, quite often, joint optimization of both system throughput and user fairness is required to make optimized decisions in the face of trade-offs between these two equally important but conflicting objectives. In this regard, this paper formulates a multi-objective optimization problem to jointly maximize the sum rate and user fairness in a downlink transmission NOMA system, through optimize power allocation (PA), under system-imposed constraints. A weighted sum approach is used to turn the multi-objective optimization problem into a single-objective optimization problem to make it analytically tractable. The optimized PA is then obtained using Lagrange dual decomposition method and Karush–Kuhn–Tucker (KKT) conditions. Using our derived expressions, we propose an iterative PA algorithm that converges fast enough to be employed in practical NOMA networks. We also present simulation results to highlight the effectiveness of the proposed solution. Further, the performance of the proposed method is compared with that of the benchmark methods.

Efficient and Secure Blockchain-Based Access Control for Fog-Assisted IoT Cloud in Electronic Medical Records Sharing

Fog-assisted IoT Cloud environments have emerged as a result of the rapid spread of Internet of Things (IoT) devices and the requirement for seamless data sharing in the healthcare industry. The effective and safe management of access control to sensitive data, such as Electronic Medical Records (EMRs), becomes crucial in such configurations. The Efficient and Secure Blockchain-Based Access Control (ESBAC) framework, designed specifically for Fog-assisted IoT Cloud scenarios in EMR sharing, is our creative approach to this problem. The ESBAC architecture makes use of the blockchain's openness and immutability to provide reliable access control and auditability. The implementation of a consortium blockchain includes participants including healthcare providers, patients, and authorised entities. Each EMR has a smart contract attached to it that specifies access restrictions and permissions to make sure that only authorised parties are able to access and edit the data. Because of the decentralisation provided by the blockchain, security and privacy are improved because no single party has overwhelming control. Proposed framework uses attribute-based access control and lightweight encryption methods to provide a fair trade-off between security and efficiency. As a result, there can be fine-grained access control policies with the least amount of computing overhead. Comprehensive simulations are used to evaluate the suggested solution and compare it to current methods. The outcomes show that ESBAC improves security in EMR sharing and greatly lowers access latency.

Chasing Fairness in Graphs: A GNN Architecture Perspective

There has been significant progress in improving the performance of graph neural networks (GNNs) through enhancements in graph data, model architecture design, and training strategies. For fairness in graphs, recent studies achieve fair representations and predictions through either graph data pre-processing (e.g., node feature masking, and topology rewiring) or fair training strategies (e.g., regularization, adversarial debiasing, and fair contrastive learning). How to achieve fairness in graphs from the model architecture perspective is less explored. More importantly, GNNs exhibit worse fairness performance compared to multilayer perception since their model architecture (i.e., neighbor aggregation) amplifies biases. To this end, we aim to achieve fairness via a new GNN architecture. We propose Fair Message Passing (FMP) designed within a unified optimization framework for GNNs. Notably, FMP explicitly renders sensitive attribute usage in forward propagation for node classification task using cross-entropy loss without data pre-processing. In FMP, the aggregation is first adopted to utilize neighbors' information and then the bias mitigation step explicitly pushes demographic group node presentation centers together. In this way, FMP scheme can aggregate useful information from neighbors and mitigate bias to achieve better fairness and prediction tradeoff performance. Experiments on node classification tasks demonstrate that the proposed FMP outperforms several baselines in terms of fairness and accuracy on three real-world datasets. The code is available at https://github.com/zhimengj0326/FMP.

Multi-objective day-ahead scheduling of cascade hydropower-photovoltaic complementary system with pumping installation

Cascade hydropower (CHP) is a promising resource to compensate for the randomness and variability of photovoltaic (PV) power generation. However, the flexibility of CHP might become insufficient due to increasing PV penetration. By constructing pump units to transform into mixed pumped-storage plants, the regulating flexibility can be further improved. Nevertheless, since the pump is usually installed between two adjacent reservoirs, the hydraulic coupling between CHP and pumps is strengthened, and it challenges the coordinated power dispatch of the CHP-PV complementary system with pumping installation. In this paper, a day-ahead dispatch strategy for the complementary system is established based on multi-objective stochastic optimization, which aims to reduce the variance of residual load, output deviation of complementary system and PV curtailment caused by PV uncertainties simultaneously. The normalized normal constraint method is employed to generate evenly distributed points on the Pareto frontier, which can present a fair tradeoff among multi-objectives. Hence, subjectivity arising from weight assignment can be eliminated and comprehensive decision support can be provided for the dispatch of the complementary system. Simulation studies are conducted on a real system in southwest China and the effectiveness of the proposed approach is validated. Compared to the case without pumping installation, the capability to compensate for PV fluctuations and the peak-shaving performance can be improved.

Group efficiency and individual fairness tradeoff in making wise decisions

In group decision making, the consensus model with minimum cost has been researched with the aim of improving group efficiency and saving resources. However, one limitation of the minimum cost consensus model is that the reach of consensus is usually at the expense of some group members. We consider two issues that we see as keys in group consensus: efficiency and fairness. We propose the price of fairness in the opinion revision process and give two kinds of fairness schemes. According to the individual's perception of inequity, we introduce inequity aversion parameters and classify experts into two types: experts with non-cooperative behaviors and with altruistic behaviors. Experts with altruistic behaviors will be allowed to contribute more than the recommended number of modifications. Then, we discuss how to achieve the tradeoff between efficiency and fairness. Furthermore, with the rapid development of social media, cloud, and e-government platforms, collective intelligence (CI), i.e., groups of individuals doing things collectively that seem intelligent, has been a hot topic. We expand our work to a crowd context with many individuals. We investigate how the opinion revision process and fairness schemes can influence the emergence of CI. Results suggest that the proportional fairness and max-min fairness have similar performance in stimulating CI. Moreover, the improvement of group accuracy is mainly related to two factors: the group consensus level of initial opinions and the relative distance between group aggregated opinion and the ground truth.

Efficiency and fairness trade-offs in two player bargaining games

Recent work on the evolution of social contracts and conventions has often used models of bargaining games, with reinforcement learning. A recent innovation is the requirement that every strategy must be invented either through through learning or reinforcement. However, agents frequently get stuck in highly-reinforced “traps” that prevent them from arriving at outcomes that are efficient or fair to the both players. Agents face a trade-off between exploration and exploitation, i.e. between continuing to invent new strategies and reinforcing strategies that have already become highly reinforced by yielding high rewards. In this paper I systematically study the relationship between rates of invention and the efficiency and fairness of outcomes in two-player, repeated bargaining games. I use a basic reinforcement learning model with invention, and five variations of this model, designed introduce various forms of forgetting, to prioritize more recent reinforcement, or to maintain a higher rate of invention. I use computer simulations to investigate the outcomes of each model. Each models shows qualitative similarities in the relationship between the efficiency and fairness of outcomes, and the relative amount of exploration or exploitation that takes place. Surprisingly, there are often trade-offs between the efficiency and the fairness of the outcomes.

Balanar: Balancing deadline guarantee and Jain’s fairness for inter-datacenter transfers

With the rising quality of service (QoS) requirements on network transmission, fairness and deadline guarantee have become the two most critical indicators on the rationality of network resource allocation and the timeliness of network transmission. However, deadline guarantee and fairness are conflicting objectives that cannot be ensured simultaneously. More specifically, preemptive scheduling policies ensure more inter-datacenter transfers can be finished before their deadlines but cause starvation, while fairly sharing network bandwidth among inter-datacenter transfers ensures fairness but cause deadlines missing. Although the state-of-the-art works have considered fairness and deadline guarantee simultaneously, a mechanism that can flexibly balance these two conflicting objectives is still missing. In this paper, we propose a decentralized traffic scheduling mechanism called Balanar. It achieves flexible trade-offs between long-term Jain’s fairness and utility of guaranteeing inter-datacenter transfers’ deadlines by finding the On-line Optimal Utility-Jain’s Fairness Trade-off (O2UJT) bandwidth allocation profiles, which maximize utility while ensuring long-term Jain’s fairness is higher than a threshold. However, finding the O2UJT bandwidth allocation profiles corresponds to solving a family of complex piece-wise non-convex optimization problems. Thus, we propose a dynamic particle swarm based heuristic algorithm to alleviate this difficulty. Finally, experimental results show Balanar can achieve flexible trade-offs between fairness and utility of guaranteeing transfers’ deadlines without damaging the other network performance indicators.

Navigating fairness measures and trade-offs

To monitor and prevent bias in AI systems, we can use a wide range of (statistical) fairness measures. However, it is mathematically impossible to optimize all of these measures at the same time. In addition, optimizing a fairness measure often greatly reduces the accuracy of the system (Kozodoi et al., Eur J Oper Res 297:1083–1094, 2022). As a result, we need a substantive theory that informs us how to make these decisions and for what reasons. I show that by using Rawls’ notion of justice as fairness, we can create a basis for navigating fairness measures and the accuracy trade-off. In particular, this leads to a principled choice focusing on both the most vulnerable groups and the type of fairness measure that has the biggest impact on that group. This also helps to close part of the gap between philosophical accounts of distributive justice and the fairness literature that has been observed by (Kuppler et al. Distributive justice and fairness metrics in automated decision-making: How much overlap is there? arXiv preprint arXiv:2105.01441, 2021), and to operationalise the value of fairness.

Optimizing of Traffic-Signal Timing Based on the FCIC-PI—A Surrogate Measure for Fuel Consumption

Optimizing signal timing improves sustainability metrics (e.g., fuel consumption or “FC”). Historically, traffic agencies have retimed signal timing to improve mobility measures (e.g., delays). However, optimizing signals to reduce delays does not necessarily mitigate sustainability measures. Hence, this study introduces an approach that integrates a newly derived surrogate measure for FC, traffic microsimulation software, and a stochastic genetic algorithm. This approach optimizes signal timing to reduce the surrogate measure and reduce sustainability metrics. This study also evaluated the impact of heavy vehicles’ presence in a fleet on signal timing and FC savings. A 13-intersection arterial on Washington Street in the Chicago metro area served as a case study. Optimized signal timing delivered solutions that balanced both sustainability and mobility. The estimated excess FC savings ranged between 8 and 12% under moderate operating conditions, with no heavy vehicles, compared to the initial signal timing. The savings reached up to ~14% when many heavy vehicles existed on the side streets. Most of the improvements came without worsening traffic-mobility efficiency, which shows the possibility of a fair tradeoff between mobility and sustainability. All optimization scenarios showed that a slightly longer cycle length than the one implemented in the field is required to reduce FC.

Deep Reinforcement Learning-Based Sum Rate Fairness Trade-Off for Cell-Free mMIMO

The uplink of a cell-free massive multiple-input multiple-output with maximum-ratio combining (MRC) and zero-forcing (ZF) schemes are investigated. A power allocation optimization problem is considered, where two conflicting metrics, namely the sum rate and fairness, are jointly optimized. As there is no closed-form expression for the achievable rate in terms of the large scale-fading (LSF) components, the sum rate fairness trade-off optimization problem cannot be solved by using known convex optimization methods. To alleviate this problem, we propose two new approaches. For the first approach, a use-and-then-forget scheme is utilized to derive a closed-form expression for the achievable rate. Then, the fairness optimization problem is iteratively solved through the proposed sequential convex approximation (SCA) scheme. For the second approach, we exploit LSF coefficients as inputs of a twin delayed deep deterministic policy gradient (TD3), which efficiently solves the non-convex sum rate fairness trade-off optimization problem. Next, the complexity and convergence properties of the proposed schemes are analyzed. Numerical results demonstrate the superiority of the proposed approaches over conventional power control algorithms in terms of the sum rate and minimum user rate for both the ZF and MRC receivers. Moreover, the proposed TD3-based power control achieves better performance than the proposed SCA-based approach as well as the fractional power scheme.

Revisiting the Adversarial Robustness-Accuracy Tradeoff in Robot Learning

Adversarial training (i.e., training on adversarially perturbed input data) is a well-studied method for making neural networks robust to potential adversarial attacks during inference. However, the improved robustness does not come for free but rather is accompanied by a decrease in overall model accuracy and performance. Recent work has shown that, in practical robot learning applications, the effects of adversarial training do not pose a fair trade-off but inflict a net loss when measured in holistic robot performance. This work revisits the robustness-accuracy trade-off in robot learning by systematically analyzing if recent advances in robust training methods and theory in conjunction with adversarial robot learning, are capable of making adversarial training suitable for real-world robot applications. We evaluate three different robot learning tasks ranging from autonomous driving in a high-fidelity environment amenable to sim-to-real deployment to mobile robot navigation and gesture recognition. Our results demonstrate that, while these techniques make incremental improvements on the trade-off on a relative scale, the negative impact on the nominal accuracy caused by adversarial training still outweighs the improved robustness by an order of magnitude. We conclude that although progress is happening, further advances in robust learning methods are necessary before they can benefit robot learning tasks in practice.

Resource Allocation in C-V2X Mode 3 Based on the Exchanged Preference Profiles

In this paper, we investigate the resource block (RB) allocation problem in cellular vehicle-to-everything (C-V2X) networks mode 3, where the cellular networks schedule the RBs for direct vehicular communications. First, we establish the communication model and introduce the effective capacity and queuing theory to describe the reliability of vehicle-to-vehicle (V2V) links. Then, we introduce the α-fair function and formulate the joint power control and RB allocation problem considering the allocation fairness and the different quality-of-service (QoS) requirements for vehicle-to-infrastructure (V2I) and V2V links. Our objective is to maximize the sum capacity of all V2I links with the α-fair function while guaranteeing the allocation fairness among V2I links and the transmission reliability for each V2V pair. To achieve this objective, we propose a novel matching game theory algorithm based on the exchanged preference profiles between the two participant sets, i.e., V2V and V2I links. Simulation results show that our proposed algorithm is adaptive to the dynamic vehicular network and achieves better efficiency and fairness trade-offs, outperforming the classic allocation method.

EASY-FIA: A Readably Usable Standalone Tool for High-Resolution Mass Spectrometry Metabolomics Data Pre-Processing.

Flow injection analysis coupled with high-resolution mass spectrometry (FIA-HRMS) is a fair trade-off between resolution and speed. However, free software available for data pre-processing is few, web-based, and often requires advanced user specialization. These tools rarely embedded blank and noise evaluation strategies, and direct feature annotation. We developed EASY-FIA, a free standalone application that can be employed for FIA-HRMS metabolomic data pre-processing by users with no bioinformatics/programming skills. We validated the tool's performance and applicability in two clinical metabolomics case studies. The main functions of our application are blank subtraction, alignment of the metabolites, and direct feature annotation by means of the Human Metabolome Database (HMDB) using a minimum number of mass spectrometry parameters. In a scenario where FIA-HRMS is increasingly recognized as a reliable strategy for fast metabolomics analysis, EASY-FIA could become a standardized and feasible tool easily usable by all scientists dealing with MS-based metabolomics. EASY-FIA was implemented in MATLAB with the App Designer tool and it is freely available for download.

Toward Enabling Network Slice Mobility to Support 6G System

Even a wider set of highly critical and latency-sensitive applications with resource needs from the access network and the edge will be supported by the 6G networks. Therefore, the 6G network will deal with diversification of service platforms. Optimizing the resource consumption of network slicing on top of a shared infrastructure will become essential to keep the operating costs on an acceptable level. Each vertical, e.g., eMBBPlus, BigCom, holographic and tactile communications can run on top of network slice with specific KPIs. Different verticals can have contradicting requirements running on top of the same infrastructure. This paper investigates the orchestration of network services within a federated end-to-end network slice, which may span over multiple cloud domains as expected to be a common scenario in 6G deployments. We introduce three optimization solutions that consider two conflicting objectives, the end-to-end delay and service relocation, for orchestrating network slice. While the first solution optimizes the end-to-end delay, the second solution optimizes the service relocation. Meanwhile, the third solution leverages the bargaining game theory for achieving optimal Pareto fair trade-off configuration to optimize both objectives. The simulation results demonstrate the efficiency of the proposed solutions to achieve their main design goals

Throughput fairness trade-offs for downlink non-orthogonal multiple access systems in 5G networks

In this work, user pairing and power allocation are proposed as a hybrid scheme to maximise throughput and achieve system fairness in the non-orthogonal multiple access (NOMA) system in 5G networks. The proposed approach is designed to improve the throughput and fairness performance of the downlink NOMA system in 5G networks. User pairing and power allocation schemes are separated to reduce resource allocation complexity. Integer linear programming is applied to perform user pairing, and particle swarm optimisation is used for power allocation. Moreover, the optimisation problem is formulated by converting multi-objective functions into a single function using the scalarisation of multi-objective optimisation problems, and the penalty function is used to prevent optimisation from violating the power, fairness, and data rate constraints. Simulation results show that the proposed model outperforms the conventional numerical approach by at least 9% of throughput maximisation and achieves an acceptable fairness rate.

Modified Adaptive Mechanism for Optimising IEEE 802.15.4 WPANs for Wireless Sensor Networks

Different applications of wireless sensor networks (WSNs) have different expectations from the working of medium access control protocols. Some value reliability more than delay incurred while some demand a fair trade-off for the factors like: Throughput, bit error rate etc. This paper evaluates the performance of wireless personal area networks from 802.15.4 group for WSNs with modified algorithm which helps in reducing the medium access delay and delay in reaching of the packet from one end to another end. In this paper certain modifications to existing algorithm have been proposed for reducing the medium access delay and to reduce the number of packets dropped. The result comparisons on the performance parameters like: network output load, generated acknowledged traffic, media access delay, battery consumed and delay in packet transmission from one end to another end that the backoff number and exponent values used for transmission play vital role for improving the performance of WSNs as they directly affect the number of packets dropped, successfully acknowledged and medium access delay.

Emergency Relief Chain for Natural Disaster Response Based on Government-Enterprise Coordination.

Public health and effective risk response cannot be promoted without a coordinated emergency process during a natural disaster. One primary problem with the emergency relief chain is the homogeneous layout of rescue organizations and reserves. There is a need for government-enterprise coordination to enhance the systemic resilience and demand orientation. Therefore, a bi-level multi-phase emergency plan model involving procurement, prepositioning and allocation is proposed. The tradeoff of efficiency, economy and fairness is offered through the multi-objective cellular genetic algorithm (MOCGA). The flood emergency in Hunan Province, China is used as a case study. The impact of multi-objective and coordination mechanisms on the relief chain is discussed. The results show that there is a significant boundary condition for the coordinated location strategy of emergency facilities and that further government coordination over the transition phase can generate optimal relief benefits. Demand orientation is addressed by the proposed model and MOCGA, with the realization of the process coordination in multiple reserves, optimal layout, and transition allocation. The emergency relief chain based on government-enterprise coordination that adapts to the evolution of disasters can provide positive actions for integrated precaution and health security.