arXiv subjects

Jianlin Jiang

Publications and source records attributed to Jianlin Jiang.

5 recordsLinked to original sources

arXiv2026

Sparse-Dense Flight Copy-Based Interactive Mechanism for Airline Integrated Recovery

Flight recovery, aircraft rerouting, and passenger reallocation are critical in airline recovery. To preserve their interdependence that is neglected by the regular sequential recovery, we consider these recovery phases from an integration perspective. In addition, we incorporate cruise speed control to enhance the recovery performance. While using flight copies is a common modelling method in airline disruption management, the resulting integrated mathematical model is challenging to solve in real time due to the large number of flight copies, especially when considering cruise speed control. This paper introduces a new sparse-dense flight copy approach and proposes an innovative interactive mechanism that alternately adjusts aircraft routes on the sparse flight copy-based network and reallocates passenger itineraries on the dense flight copy-based network. Under the interactive mechanism, the involved sparse and dense networks are much smaller than those in the conventional flight copy approach. To implement such a mechanism, we develop an integrated flight, aircraft, and passenger recovery model (IFAPRM) and propose a customized Benders decomposition (CBD) to solve the model. Besides, we further propose some acceleration techniques to speed up the CBD method, including an effective feasibility certificate, scale management, and valid inequalities. Computational experiments on real-world data demonstrate that the sparse-dense flight copy-based interactive mechanism outperforms the conventional flight copy approach. In essence, the proposed interactive mechanism, along with its corresponding modelling method, algorithm, and acceleration techniques, provides a comprehensive methodology and a general decision-support framework for integrated rescheduling problems in complex operations, with potential applications in logistics, transportation, and beyond.

math.OC↗

arXiv2025

A Benders and column generation method to the integrated airline schedule and aircraft recovery with gate reassignment

Disruptions are inevitable during airline operations, and disruptions cost airlines and the traveling public. Disruption recovery decisions are often made in a sequential process that includes flight rescheduling, aircraft rerouting, crew reassignment, passenger re-accommodation, and gate reassignment for airline operations. Such a sequential recovery approach alleviates the complexity of the whole recovery process but often causes further disruptions to airport gate assignments and high recovery costs for airlines and passengers. This paper presents a novel airline disruption recovery approach by integrating the schedule and aircraft recovery with gate reassignment. We propose a Benders and column generation (BCG) method to solve the integrated problem. By exploiting the favorable structure arising from the Benders decomposition framework, we provide two acceleration techniques for Benders subproblems, a separation technique and an effective infeasibility certificate, to enhance the efficiency of the BCG method. The proposed model is tested on real-world data. In our experiments, all test instances are solved by the BCG method under a five-minute threshold with optimality gaps within 5%. Compared with the sequential recovery approach, the integrated recovery approach significantly saves recovery costs and avoids infeasible gate reassignments caused by the sequential recovery approach.

math.OC↗

arXiv2020

Shielding Collaborative Learning: Mitigating Poisoning Attacks through Client-Side Detection

Collaborative learning allows multiple clients to train a joint model without sharing their data with each other. Each client performs training locally and then submits the model updates to a central server for aggregation. Since the server has no visibility into the process of generating the updates, collaborative learning is vulnerable to poisoning attacks where a malicious client can generate a poisoned update to introduce backdoor functionality to the joint model. The existing solutions for detecting poisoned updates, however, fail to defend against the recently proposed attacks, especially in the non-IID setting. In this paper, we present a novel defense scheme to detect anomalous updates in both IID and non-IID settings. Our key idea is to realize client-side cross-validation, where each update is evaluated over other clients' local data. The server will adjust the weights of the updates based on the evaluation results when performing aggregation. To adapt to the unbalanced distribution of data in the non-IID setting, a dynamic client allocation mechanism is designed to assign detection tasks to the most suitable clients. During the detection process, we also protect the client-level privacy to prevent malicious clients from stealing the training data of other clients, by integrating differential privacy with our design without degrading the detection performance. Our experimental evaluations on two real-world datasets show that our scheme is significantly robust to two representative poisoning attacks.

cs.CR↗

arXiv2019

Estimating and decomposing most productive scale size in parallel DEA networks with shared inputs: A case of China's Five-Year Plans

Attaining the optimal scale size of production systems is an issue frequently found in the priority questions on management agendas of various types of organizations. Determining the most productive scale size (MPSS) allows the decision makers not only to know the best scale size that their systems can achieve but also to tell the decision makers how to move the inefficient systems onto the MPSS region. This paper investigates the MPSS concept for production systems consisting of multiple subsystems connected in parallel. First, we propose a relational model where the MPSS of the whole system and the internal subsystems are measured in a single DEA implementation. Then, it is proved that the MPSS of the system can be decomposed as the weighted sum of the MPSS of the individual subsystems. The main result is that the system is overall MPSS if and only if it is MPSS in each subsystem. MPSS decomposition allows the decision makers to target the non-MPSS subsystems so that the necessary improvements can be readily suggested. An application of China's Five-Year Plans (FYPs) with shared inputs is used to show the applicability of the proposed model for estimating and decomposing MPSS in parallel network DEA. Industry and Agriculture sectors are selected as two parallel subsystems in the FYPs. Interesting findings have been noticed. Using the same amount of resources, the Industry sector had a better economic scale than the Agriculture sector. Furthermore, the last two FYPs, 11th and 12th, were the perfect two FYPs among the others.

econ.GN↗

arXiv2019

Most productive scale size of China's regional R&D value chain: A mixed structure network

This paper offers new mathematical models to measure the most productive scale size (MPSS) of production systems with mixed structure networks (mixed of series and parallel). In the first property, we deal with a general multi-stage network which can be transformed, using dummy processes, into a series of parallel networks. In the second property, we consider a direct network combined with series and parallel structure. In this paper, we propose new models to measure the overall MPSS of the production systems and their internal processes. MPSS decomposition is discussed and examined. As a real-life application, this study measures the efficiency and MPSS of research and development (R&D) activities of Chinese provinces within an R&D value chain network. In the R&D value chain, profitability and marketability stages are connected in series, where the profitability stage is composed of operation and R&D efforts connected in parallel. The MPSS network model provides not only the MPSS measurement but also values that indicate the appropriate degree of intermediate measures for the two stages. Improvement strategy is given for each region based on the gap between the current and the appropriate level of intermediate measures. Our findings show that the marketability efficiency values of Chinese R&D regions were low, and no regions are operated under the MPSS. As a result, most Chinese regions performed inefficiently regarding both profitability and marketability. This finding provides initial evidence that the generally lower profitability and marketability efficiency of Chinese regions is a severe problem that may be due to wasted resources on production and R&D.

econ.GN↗