APAndrey Poddubnyy

Andrey Poddubnyy

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Andrey Poddubnyy has a total of 13 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Addressing the Collaboration Dilemma in Low-Data Federated Learning via Transient Sparsity

    by Qiao Xiao, Boqian Wu, Andrey Poddubnyy, Elena Mocanu, Phuong H. Nguyen, Mykola Pechenizkiy, Decebal Constantin Mocanu

    Federated learning (FL) enables collaborative model training across decentralized clients while preserving data privacy, leveraging aggregated updates to build robust global models. However, this training paradigm faces significant challenges due to data heterogeneity and limited local datasets, which often impede effective collaboration. In such scenarios, we identify the Layer-wise Inertia Phenomenon in FL, wherein the middle layers of global model undergo minimal updates after early communication rounds, ultimately limiting the effectiveness of global aggregation. We demonstrate the presence of this phenomenon across a wide range of federated settings, spanning diverse datasets and architectures. To address this issue, we propose LIPS (Layer-wise Inertia Phenomenon with Sparsity), a simple yet effective method that periodically introduces transient sparsity to stimulate meaningful updates and empower global aggregation. Experiments demonstrate that LIPS effectively mitigates layer-wise inertia, enhances aggregation effectiveness, and improves overall performance in various FL scenarios. This work not only deepens the understanding of layer-wise learning dynamics in FL but also paves the way for more effective collaboration strategies in resource-constrained environments. Our code is publicly available at: https://github.com/QiaoXiao7282/LIPS.

  • Open Access

    Coordinated Optimization of Logistics Electric Fleet and Energy Management System of Constrained Energy Hub

    by Ali Saklaoui, Andrey Poddubnyy, Phuong Nguyen, Christina Papadimitriou, Quang-Vinh Dang, Rajiv Hotchandani

    The electricity network has reached its transport capacity limits in various areas in the Netherlands and the challenge of granting connections became critical. The concept of energy hubs, where neighboring prosumers collaborate to optimize the available capacity, poses itself as a short-term alternative for grid reinforcement. This study presents a coordinated, MPC-based combined with the partheno-genetic algorithm, optimization approach enabling a smooth transition into an electric fleet for logistics companies, taking part of an energy hub, while respecting the grid's limited capacity and taking into account uncertainties arising from load and generation profiles. The last-mile deliveries of the logistic company are depicted by the Electric Vehicle Routing Problem formulation, and the partheno-genetic algorithm is implemented to solve it, where the parameters of interest are fed to the energy management system that minimizes the overall energy costs at the energy hub while incorporating day-ahead market prices. The intermittency of renewable generation and load demand is tackled by adopting the model predictive control (MPC) framework, providing a corrective mechanism that ensures that the overarching objective of the system is met while respecting the grid's limitation. Three charging strategies at the hub are investigated: dynamic charging, where charging power varies by magnitude and time, direct charging, having a fixed charging power and time, and delayed overnight charging, where the charging power is spread over a scheduled horizon. The results demonstrate the mitigation of the grid's connection capacity while attaining cost savings with the proposed dynamic charging strategy, ranging between 11.5% and 52% compared to the delayed overnight charging and direct charging strategies.


  • Network-Aware Online Charge Control with Reinforcement Learning

    by Andrey Poddubnyy, Phuong Nguyen, Han Slootweg

    The rising market of electric vehicles leads to an increased number of chargers needed to be connected to the power grid. It sometimes leads to their installation being postponed due to the capacity limits violations, which can be eased by implementing a smart charging control. In this paper the smart charging control is proposed from the point of view of grid loading as a primary indicator. The reinforcement learning agent controls the charger’s power of consumption to optimize expenses and prevent lines and transformers from being overloaded. The simulations were carried out in the IEEE 13 bus test feeder with the load profile data coming from the residential area. To simulate the real availability of data, an agent is trained with only the transformer current and the local charger’s state, like state of the charge and timestamp. Several algorithms are tested to select the best one to utilize in the stochastic environment and low frequency of data streaming.