Pith. sign in

REVIEW 2 major objections 1 minor 66 references

Explainable Data-driven Deep Reinforcement Learning Methods for Optimal Energy Management in Buildings

T0 review · 2 major / 1 minor · reviewed 2026-06-28 · grok-4.3

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read On-policy reinforcement learning agents for building energy management outperform off-policy methods and can be made interpretable through post-hoc analysis of their battery control decisions.

desk verdict This applies standard on-policy DRL (A2C/PPO) plus post-hoc explainers to building energy control on real LLEC data, but the performance claims lack error bars, tests, or a rule-based baseline. read the letter →

arxiv 2606.02049 v1 pith:MPEOS4BC submitted 2026-06-01 cs.AI

Hallah Shahid Butt , Qiong Huang , Gökhan Demirel , Kevin Förderer , Erfan Tajalli-Ardekani , Simnon Waczowicz , Luigi Spatafora , Veit Hagenmeyer
show 1 more author
Benjamin Schäfer
This is my paper
classification cs.AI
keywords explainablereinforcementlearningenergymanagementdeepbuildingsystemspost-hocinterpretationon-policyalgorithmsbatteryresidentialbuildings
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces an explainable reinforcement learning framework for optimizing energy use in residential buildings that have solar panels, batteries, and variable electricity prices. Agents are trained on an expanded state space that includes real-time measurements, forecasts, weather, prices, and calendar data from both synthetic and real Living Lab sources. On-policy algorithms, especially A2C and PPO, produce higher cumulative rewards and more stable policies than off-policy alternatives. Post-hoc interpretation methods are then applied to surface the rules the agents follow when deciding when to charge or discharge the battery. The result is both lower electricity costs and explicit, human-readable explanations of the control actions.

What carries the argument

Post-hoc interpretation techniques applied to Advantage Actor Critic (A2C) and Proximal Policy Optimization (PPO) agents trained on an expanded state space containing demand, PV generation, battery state, dynamic prices, weather, and forecasts.

What would settle it

A controlled test in which the extracted explanations are used to predict the agent's battery charge/discharge actions on held-out scenarios but the predictions systematically mismatch the actual policy outputs.

Watch

Extended reading notes

Core claim

The paper establishes that on-policy deep reinforcement learning agents trained for residential energy management achieve superior cumulative rewards and policy stability compared with off-policy agents, and that post-hoc interpretation techniques applied to these agents yield transparent, actionable descriptions of the learned battery management policy.

Load-bearing premise

Post-hoc interpretation techniques applied after training faithfully capture the causal drivers of the learned policy rather than merely correlating with observed actions.

Editorial extensions

If this is right

  • On-policy methods yield more stable control policies than off-policy methods when electricity prices and demand vary over time.
  • Optimal battery scheduling learned by these agents measurably lowers electricity costs in buildings equipped with PV and storage.
  • Actionable explanations of the learned policy increase the chance that building operators will trust and adopt the controller.
  • The same training and explanation pipeline can be retrained on new building data without changing the overall architecture.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the explanations prove reliable, the same post-hoc pipeline could be reused to audit RL controllers for other building subsystems such as heat pumps or EV chargers.
  • The performance gap between on-policy and off-policy agents may widen further when the state space includes additional stochastic elements such as occupant behavior.
  • Transparent battery policies could be converted into simple rule-based controllers for deployment on low-cost hardware.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 1 minor

Summary. The manuscript presents an explainable deep reinforcement learning (XRL) framework for optimal energy management in residential buildings equipped with PV and battery storage. It expands the state space with real-time measurements, dynamic prices, weather, calendars, and forecasts; trains and compares on-policy (A2C, PPO) versus off-policy DRL agents on both synthetic data and real LLEC measurements; reports that on-policy methods achieve higher cumulative rewards and greater stability; and applies post-hoc interpretation techniques to generate transparent insights into the learned battery-control policies that reduce electricity costs.

Significance. If the superiority and explanatory claims hold after proper statistical validation, the work would strengthen the case for deploying DRL in building energy systems by addressing the black-box barrier. The use of real LLEC data alongside synthetic cases is a concrete strength that grounds the results in operational conditions.

major comments (2)
  1. [Abstract] Abstract and the experimental-results paragraph: the claim that 'on-policy algorithms, particularly A2C and PPO, outperform off-policy methods in terms of cumulative rewards and policy stability' is presented without error bars, statistical significance tests, a rule-based baseline controller, or any description of hyper-parameter search or train/test splits. These omissions make the central superiority assertion impossible to evaluate from the reported data.
  2. [XRL framework description] The paragraph describing the XRL framework and post-hoc techniques: the assertion that the methods 'provide transparent, actionable insights into the agent's decision-making process' rests on post-hoc attribution applied after training. No ablation, intervention, or counterfactual test is described that would establish whether the highlighted features causally drive the policy outputs rather than merely correlating with observed actions in a non-stationary environment containing forecasts and prices.
minor comments (1)
  1. [Abstract] The abstract mentions 'an expanded state space' but does not list the exact feature dimensions or normalization procedures used; adding this detail would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and indicate planned revisions to improve statistical rigor and clarify the scope of the explanatory claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract and the experimental-results paragraph: the claim that 'on-policy algorithms, particularly A2C and PPO, outperform off-policy methods in terms of cumulative rewards and policy stability' is presented without error bars, statistical significance tests, a rule-based baseline controller, or any description of hyper-parameter search or train/test splits. These omissions make the central superiority assertion impossible to evaluate from the reported data.

    Authors: We agree that the current presentation of results lacks the statistical details needed for rigorous evaluation. In the revised manuscript we will report error bars computed over multiple independent training runs with different random seeds, include statistical significance tests (paired t-tests or Wilcoxon tests) on cumulative rewards, add a simple rule-based baseline controller (e.g., SOC-threshold logic), and expand the methods section with the hyper-parameter search procedure and the train/test split protocol used for both synthetic and LLEC data. These additions will directly address the evaluability concern. revision: yes

  2. Referee: [XRL framework description] The paragraph describing the XRL framework and post-hoc techniques: the assertion that the methods 'provide transparent, actionable insights into the agent's decision-making process' rests on post-hoc attribution applied after training. No ablation, intervention, or counterfactual test is described that would establish whether the highlighted features causally drive the policy outputs rather than merely correlating with observed actions in a non-stationary environment containing forecasts and prices.

    Authors: We acknowledge that post-hoc attribution yields correlational rather than causal evidence. The manuscript employs standard techniques (e.g., SHAP values) to surface feature importance for the learned policies; this is the conventional approach in current XRL work. We will revise the relevant paragraph to explicitly state that the insights are correlational, add a limitations subsection discussing the absence of ablation or counterfactual experiments, and, if space allows, include a brief feature-ablation check on the most salient attributions. Full causal intervention studies would require new experimental infrastructure and are noted as future work. revision: partial

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical RL results on external LLEC data

full rationale

The paper trains and evaluates standard DRL agents (A2C, PPO, etc.) on real-world measurements from the external LLEC dataset, comparing cumulative rewards and stability. Post-hoc explanation methods are applied after training but introduce no derivation that reduces outputs to self-defined inputs or fitted parameters. No equations, self-citations, or ansatzes are shown to make the central claims equivalent to their own inputs by construction. The evaluation chain is self-contained against external benchmarks.

Assumptions & free parameters 0 free parameters · 0 assumptions · 0 invented entities

Paper is an empirical application study; no new mathematical axioms, free parameters fitted inside the central claim, or invented physical entities are introduced in the abstract.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Explainable Data-driven Deep Reinforcement Learning Methods for Optimal Energy Management in Buildings." pith.science (2026). https://pith.science/paper/MPEOS4BC

@misc{pith2026260602049,
  author       = {Pith},
  title        = {Pith review of: Explainable Data-driven Deep Reinforcement Learning Methods for Optimal Energy Management in Buildings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MPEOS4BC}},
  note         = {Machine review of arXiv:2606.02049}
}
read the original abstract

The increasing integration of renewable energy sources into power systems, particularly in buildings equipped with photovoltaic (PV) panels and energy storage systems, introduces significant complexity in energy systems. Volatile power generation, varying electricity tariffs, and increased entities, e.g., PV systems, and heat pumps, have increased the complexity and made the system harder to operate. This leads to the demand for additional control and optimization routes including data-based controls, such as reinforcement learning. While deep reinforcement learning (DRL) has emerged as a promising solution to optimize building operations in dynamic and ever more complex environments, its black-box nature impedes user trust and practical adoption. This paper presents a framework for explainable deep reinforcement learning (XRL) applied to energy management in residential buildings. We demonstrate its usage on both synthetic data but also on real-world data from the Living Lab Energy Campus (LLEC) at KIT. We train and compare both on-policy and off-policy DRL agents on an expanded state space that incorporates real-time measurements (demand, PV generation, battery power, state of charge), external signals (dynamic electricity price, local weather data), calendrical and holiday indicators, and forecasts for demand and price. Our experimental results indicate that on-policy algorithms, particularly Advantage Actor Critic (A2C) and Proximal Policy Optimization (PPO), outperform off-policy methods in terms of cumulative rewards and policy stability. To explain these models, we employ post-hoc interpretation techniques to elaborate the learned control policies. Our findings demonstrate that the XRL framework not only reduces electricity costs through optimal battery management, but also provides transparent, actionable insights into the agent's decision-making process.

Figures

Figures reproduced from arXiv: 2606.02049 by the authors.

Figure 1
Figure 1. FIGURE 1 [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIGURE 3 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: FIGURE 4 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: FIGURE 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: FIGURE 6 [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: FIGURE 7 [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: FIGURE 8 [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: FIGURE 9 [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

66 extracted references · 3 canonical work pages

  1. [1]

    Role of opti- mization techniques in microgrid energy management systems—a review

    Gokul Sidarth Thirunavukkarasu, Mehdi Seyedmahmoudian, Elmira Jamei, Ben Horan, Saad Mekhilef, and Alex Stojcevski. Role of opti- mization techniques in microgrid energy management systems—a review. Energy Strategy Reviews, 43:100899, 2022

    2022

  2. [2]

    Explainable reinforcement learning for optimizing electricity costs in building energy management

    Hallah Shahid Butt, Qiong Huang, and Benjamin Schäfer. Explainable reinforcement learning for optimizing electricity costs in building energy management. In2024 3rd International Conference on Energy Transition in the Mediterranean Area (SyNERGY MED), pages 1–6. IEEE, 2024

    2024

  3. [3]

    Optimal energy system scheduling using a constraint-aware reinforcement learning algorithm.International Journal of Electrical Power & Energy Systems, 152:109230, 2023

    Hou Shengren, Pedro P V ergara, Edgar Mauricio Salazar Duque, and Peter Palensky. Optimal energy system scheduling using a constraint-aware reinforcement learning algorithm.International Journal of Electrical Power & Energy Systems, 152:109230, 2023

    2023

  4. [4]

    https://www

    Monthly sunshine hours germany 2024-2025| statista. https://www. statista.com/statistics/982758/average-sunshine-hours-germany/. Ac- cessed: 2025-06-30

    2024

  5. [5]

    Theory and applications of hvac control systems – a review of model predictive control (mpc).Building and Environment, 72:343–355, 2014

    Abdul Afram and Farrokh Janabi-Sharifi. Theory and applications of hvac control systems – a review of model predictive control (mpc).Building and Environment, 72:343–355, 2014

    2014

  6. [6]

    Ten questions concerning model pre- dictive control for energy efficient buildings.Building and Environment, 105:403–412, 2016

    Markus Killian and Manfred Kozek. Ten questions concerning model pre- dictive control for energy efficient buildings.Building and Environment, 105:403–412, 2016

    2016

  7. [7]

    Predic- tive control for energy efficient buildings with thermal storage: Modeling, simulation, and experiments.IEEE Control Systems Magazine, 32(1):44– 64, 2012

    Y udong Ma, Adam Kelman, Aaron Daly, and Francesco Borrelli. Predic- tive control for energy efficient buildings with thermal storage: Modeling, simulation, and experiments.IEEE Control Systems Magazine, 32(1):44– 64, 2012

    2012

  8. [8]

    All you need to know about model predictive control for buildings.Annual Reviews in Control, 50:190–232, 2020

    Ján Drgoňa, Javier Arroyo, Itziar Cupeiro Figueroa, David Blum, Krzysztof Arendt, Donghun Kim, Elena Ollé, Jozef Oravec, Michael Wet- ter, Draguna Vrabie, and Lieve Helsen. All you need to know about model predictive control for buildings.Annual Reviews in Control, 50:190–232, 2020

    2020

Show all 66 references
  1. [9]

    David Sturzenegger, Dimitrios Gyalistras, Manfred Morari, and Roy S. Smith. Model predictive climate control of a swiss office building: Im- plementation, results, and cost–benefit analysis.IEEE Transactions on Control Systems Technology, 24(1):1–12, 2016

    2016

  2. [10]

    Importance of occupancy information for building climate control.Applied Energy, 101:521–532, 2013

    Frauke Oldewurtel, David Sturzenegger, and Manfred Morari. Importance of occupancy information for building climate control.Applied Energy, 101:521–532, 2013

    2013

  3. [11]

    A coordinated model predic- tive control approach for building energy management considering real- time electricity prices.Energy, 164:792–802, 2018

    Xin Zhang, Chongqing Kang, and Qing Xia. A coordinated model predic- tive control approach for building energy management considering real- time electricity prices.Energy, 164:792–802, 2018

    2018

  4. [12]

    A review of reinforcement learning for energy management in smart buildings.Energy and AI, 10:100204, 2022

    Tao Wei, Y ujie Wang, Y an Li, and Tong Zhang. A review of reinforcement learning for energy management in smart buildings.Energy and AI, 10:100204, 2022

    2022

  5. [13]

    Deep reinforcement learning for building hvac control: Review and chal- lenges.Energy and Buildings, 240:110886, 2021

    Yixuan Zhang, Hoi-Lun Lam, Dong Li, Jianjun Qin, and Ho Ting Chan. Deep reinforcement learning for building hvac control: Review and chal- lenges.Energy and Buildings, 240:110886, 2021

    2021

  6. [14]

    Norford, and Ali M

    Y ulong Chen, Leslie K. Norford, and Ali M. Malkawi. Reinforcement learning-based control for building energy management: A review and case study.Applied Energy, 281:116060, 2021

    2021

  7. [15]

    A reinforcement learning approach for optimal control of building hvac systems under partial ob- servability.Energy and Buildings, 223:110135, 2020

    Guanqi Gao, Zhiwei Shi, and Xuesong Lu. A reinforcement learning approach for optimal control of building hvac systems under partial ob- servability.Energy and Buildings, 223:110135, 2020

    2020

  8. [16]

    Model-free control of building energy systems using deep reinforcement learning: Recent progress and future prospects.Energy and Buildings, 285:112885, 2023

    Bowen Y u, Ján Drgoňa, David Blum, and Michael Wetter. Model-free control of building energy systems using deep reinforcement learning: Recent progress and future prospects.Energy and Buildings, 285:112885, 2023

    2023

  9. [17]

    Multi-objective reinforcement learning with shielding for control and safety in power grids.ACM SIGEnergy Energy Informatics Review, 5(3):93–104, 2025

    Gökhan Demirel, Simon Grafenhorst, Jakob Ohm, Benjamin Schäfer, Kevin Förderer, and V eit Hagenmeyer. Multi-objective reinforcement learning with shielding for control and safety in power grids.ACM SIGEnergy Energy Informatics Review, 5(3):93–104, 2025

    2025

  10. [18]

    Feasibility of low-cost energy management system using embedded opti- mization for pv and battery storage assisted residential buildings.Energy, 271:126922, 2023

    Kiswendsida Elias Ouedraogo, Pınar Oğuz Ekim, and Erhan Demirok. Feasibility of low-cost energy management system using embedded opti- mization for pv and battery storage assisted residential buildings.Energy, 271:126922, 2023

    2023

  11. [19]

    A re- view on optimal energy management in commercial buildings.Energies, 16(4):1609, 2023

    Jahangir Hossain, Aida FA Kadir, Ainain N Hanafi, Hussain Shareef, Tamer Khatib, Kyairul A Baharin, and Mohamad F Sulaima. A re- view on optimal energy management in commercial buildings.Energies, 16(4):1609, 2023

    2023

  12. [20]

    Internet of things applications for energy management in buildings using artificial intelligence—a case study.Energies, 18(7):1706, 2025

    Izabela Rojek, Dariusz Mikołajewski, Adam Mroziński, Marek Macko, Tomasz Bednarek, and Krzysztof Tyburek. Internet of things applications for energy management in buildings using artificial intelligence—a case study.Energies, 18(7):1706, 2025

    2025

  13. [21]

    Explaining reinforcement learning with shapley values

    Daniel Beechey, Thomas MS Smith, and Özgür Şimşek. Explaining reinforcement learning with shapley values. InInternational Conference on Machine Learning, pages 2003–2014. PMLR, 2023

    2003

  14. [22]

    Artificial intelligence and machine learning in energy systems: A biblio- graphic perspective.Energy Strategy Reviews, 45:101017, 2023

    Ashkan Entezari, Alireza Aslani, Rahim Zahedi, and Y ounes Noorollahi. Artificial intelligence and machine learning in energy systems: A biblio- graphic perspective.Energy Strategy Reviews, 45:101017, 2023

    2023

  15. [23]

    Machine learning and deep learning in energy systems: A review

    Mohammad Mahdi Forootan, Iman Larki, Rahim Zahedi, and Abolfazl Ahmadi. Machine learning and deep learning in energy systems: A review. Sustainability, 14(8), 2022

    2022

  16. [24]

    Net zero by 2050: A roadmap for the global energy sector

    International Energy Agency. Net zero by 2050: A roadmap for the global energy sector. Technical report, International Energy Agency (IEA), May

  17. [25]

    https://www.iea.org/reports/net-zero-by-2050, Accessed: 2026-04- 24

  18. [26]

    Energy efficiency in buildings 2023

    International Energy Agency. Energy efficiency in buildings 2023. Tech- nical report, International Energy Agency (IEA), November 2023. https: //www.iea.org/reports/energy-efficiency-2023, Accessed: 2026-04-24

    2023

  19. [27]

    A comprehensive review of building energy management systems: Current status and future directions.Energy and Buildings, 290:113053, 2023

    Qinglong Xu, Y u Sun, Y ang Zhang, and Shuo Wang. A comprehensive review of building energy management systems: Current status and future directions.Energy and Buildings, 290:113053, 2023

    2023

  20. [28]

    Ai-driven innovations in building energy management systems: A review of potential applications and energy savings.Energies, 17(17), 2024

    Dalia Mohammed Talat Ebrahim Ali, Violeta Motuzien ˙e, and Rasa Džiu- gait˙e-Tum˙enien˙e. Ai-driven innovations in building energy management systems: A review of potential applications and energy savings.Energies, 17(17), 2024

    2024

  21. [29]

    Reinforcement learning and its applications in modern power and energy systems: A review.Journal of modern power systems and clean energy, 8(6):1029–1042, 2020

    Di Cao, Weihao Hu, Junbo Zhao, Guozhou Zhang, Bin Zhang, Zhou Liu, Zhe Chen, and Frede Blaabjerg. Reinforcement learning and its applications in modern power and energy systems: A review.Journal of modern power systems and clean energy, 8(6):1029–1042, 2020

    2020

  22. [30]

    Advanced Deep Reinforcement Learning for Heat Pump Control in Residential Buildings

    Gökhan Demirel, Ömer Ekin, Jianlei Liu, Luigi Spatafora, Kevin Förderer, and V eit Hagenmeyer. Advanced Deep Reinforcement Learning for Heat Pump Control in Residential Buildings. In2025 IEEE PES Innovative Smart Grid Technologies Conference Europe (ISGT Europe), pages 1–5, 2025

    2025

  23. [31]

    Optimization of the operation and maintenance of renewable energy systems by deep reinforcement learning.Renewable Energy, 183:752–763, 2022

    Luca Pinciroli, Piero Baraldi, Guido Ballabio, Michele Compare, and Enrico Zio. Optimization of the operation and maintenance of renewable energy systems by deep reinforcement learning.Renewable Energy, 183:752–763, 2022

    2022

  24. [32]

    Deep reinforcement learning for power system applications: An overview.CSEE Journal of Power and Energy Systems, 6(1):213–225, 2019

    Zidong Zhang, Dongxia Zhang, and Robert C Qiu. Deep reinforcement learning for power system applications: An overview.CSEE Journal of Power and Energy Systems, 6(1):213–225, 2019

    2019

  25. [33]

    Occupant-oriented economic model predictive control for demand response in buildings

    Moritz Frahm, Philipp Zwickel, Jan Wachter, Felix Langner, Pascal Strauch, Jörg Matthes, and V eit Hagenmeyer. Occupant-oriented economic model predictive control for demand response in buildings. InProceedings of the 13th ACM International Conference on Future Energy Systems ...

    2022

  26. [34]

    Çakmak, Jörg Matthes, and V eit Hagenmeyer

    Felix Langner, Jovana Kovačević, Philipp Zwickel, Thomas Dengiz, Moritz Frahm, Simon Waczowicz, Hüseyin K. Çakmak, Jörg Matthes, and V eit Hagenmeyer. Coordinated price-based control of modulating heat pumps for practical demand response and peak shaving in building clusters. ...

    2024

  27. [35]

    A generalizable reinforcement learning framework for hvac control using real-world data.Energy and Buildings, 305:113941, 2024

    Xiaoyu Mao, Y an Cheng, Xue Li, Haoran Li, Rui Y ang, and Tianzhen Hong. A generalizable reinforcement learning framework for hvac control using real-world data.Energy and Buildings, 305:113941, 2024

    2024

  28. [36]

    Optimization strategy based on deep reinforcement learning for home energy management.CSEE Journal of Power and Energy Systems, 6(3):572–582, 2020

    Y uankun Liu, Dongxia Zhang, and Hoay Beng Gooi. Optimization strategy based on deep reinforcement learning for home energy management.CSEE Journal of Power and Energy Systems, 6(3):572–582, 2020

    2020

  29. [37]

    Comparison of reinforcement learning and model predictive control for building energy system optimization.Applied Thermal Engineering, 228:120430, 2023

    Dan Wang, Wanfu Zheng, Zhe Wang, Y aran Wang, Xiufeng Pang, and Wei Wang. Comparison of reinforcement learning and model predictive control for building energy system optimization.Applied Thermal Engineering, 228:120430, 2023

    2023

  30. [38]

    Rl4rs: A real- world dataset for reinforcement learning based recommender system

    Kai Wang, Zhene Zou, Minghao Zhao, Qilin Deng, Y ue Shang, Yile Liang, Runze Wu, Xudong Shen, Tangjie Lyu, and Changjie Fan. Rl4rs: A real- world dataset for reinforcement learning based recommender system. In Proceedings of the 46th International ACM SIGIR Conference on Resea...

    2023

  31. [39]

    A review of deep reinforcement learning for smart building energy management.IEEE Internet of Things Journal, 8(15):12046–12063, 2021

    Liang Y u, Shuqi Qin, Meng Zhang, Chao Shen, Tao Jiang, and Xiaohong Guan. A review of deep reinforcement learning for smart building energy management.IEEE Internet of Things Journal, 8(15):12046–12063, 2021

    2021

  32. [40]

    Energyplus™, 09 2017

    National Renewable Energy Laboratory (NREL). Energyplus™, 09 2017

    2017

  33. [41]

    Reviewing the need for explainable artificial intelligence (xAI).arXiv preprint arXiv:2012.01007, 2020

    Julie Gerlings, Arisa Shollo, and Ioanna Constantiou. Reviewing the need for explainable artificial intelligence (xAI).arXiv preprint arXiv:2012.01007, 2020

    2012

  34. [42]

    Toward ex- plainable and interpretable building energy modelling: An explainable VOLUME 00, 0000 13 Preprint submitted to IEEE Access artificial intelligence approach

    Wei Zhang, Fang Liu, Y onggang Wen, and Bernard Nee. Toward ex- plainable and interpretable building energy modelling: An explainable VOLUME 00, 0000 13 Preprint submitted to IEEE Access artificial intelligence approach. InProceedings of the 8th ACM Interna- tional Conference ...

    2021

  35. [43]

    An interpretable machine learning model for daily global solar radiation prediction.Energies, 14(21), 2021

    Mohamed Chaibi, EL Mahjoub Benghoulam, Lhoussaine Tarik, Mohamed Berrada, and Abdellah El Hmaidi. An interpretable machine learning model for daily global solar radiation prediction.Energies, 14(21), 2021

    2021

  36. [44]

    Ex- plainable AI (XAI) techniques for energy and power systems: Review, challenges, and future directions.Energy and AI, 9:100169, 2022

    Roy Machlev, Roee Melamed, Andrey Belikov, and Y oash Levron. Ex- plainable AI (XAI) techniques for energy and power systems: Review, challenges, and future directions.Energy and AI, 9:100169, 2022

    2022

  37. [45]

    Explainable long-term building energy consumption prediction using qlattice.Applied Energy, 308:118300, 2022

    Simon Wenninger, Can Kaymakci, and Christian Wiethe. Explainable long-term building energy consumption prediction using qlattice.Applied Energy, 308:118300, 2022

    2022

  38. [46]

    Explainable reinforcement learning: A survey

    Erika Puiutta and Eric MSP V eith. Explainable reinforcement learning: A survey. InInternational cross-domain conference for machine learning and knowledge extraction, pages 77–95. Springer, 2020

    2020

  39. [47]

    Explain- able ai framework for reliable and transparent automated energy manage- ment in buildings.Energy and Buildings, 347:116246, 2025

    Brígida Teixeira, Leonor Carvalhais, Tiago Pinto, and Zita V ale. Explain- able ai framework for reliable and transparent automated energy manage- ment in buildings.Energy and Buildings, 347:116246, 2025

    2025

  40. [48]

    https://www.iai.kit.edu/english/RPE-LLEC.php

    KIT - IAI - RPE- LLEC. https://www.iai.kit.edu/english/RPE-LLEC.php. Accessed: 2025-06-30

    2025

  41. [49]

    Electricity market transparency

    entso e. Electricity market transparency. https://www.entsoe.eu/data/ transparency-platform/, 2015

    2015

  42. [50]

    Hyndman and George Athanasopoulos.F orecasting: Principles and Practice

    Rob J. Hyndman and George Athanasopoulos.F orecasting: Principles and Practice. OTexts, 2nd edition, 2018. Accessed: 2024-01-01

    2018

  43. [51]

    Long short-term memory

    Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory. Neural Computation, 9(8):1735–1780, 1997

    1997

  44. [52]

    Rusu, Joel V eness, Marc G

    V olodymyr Mnih, Koray Kavukcuoglu, David Silver, Andrei A. Rusu, Joel V eness, Marc G. Bellemare, Alex Graves, Martin Riedmiller, Andreas K. Fidjeland, Georg Ostrovski, Stig Petersen, Charles Beattie, Amir Sadik, Ioannis Antonoglou, Helen King, Dharshan Kumaran, Daan Wierstra...

    2015

  45. [53]

    Proximal policy optimization algorithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. Proximal policy optimization algorithms. InProceedings of the 34th International Conference on Machine Learning (ICML), pages 1–12, 2017

    2017

  46. [54]

    Asynchronous methods for deep reinforcement learning

    V olodymyr Mnih, Adria Puigdomenech Badia, Mehdi Mirza, Alex Graves, Timothy Lillicrap, Tim Harley, David Silver, and Koray Kavukcuoglu. Asynchronous methods for deep reinforcement learning. InProceedings of the 33rd International Conference on Machine Learning (ICML), pages 1...

    1928

  47. [55]

    Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor

    Tuomas Haarnoja, Aurick Zhou, Pieter Abbeel, and Sergey Levine. Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. InProceedings of the 35th International Conference on Machine Learning (ICML), pages 1856–1865, 2018

    2018

  48. [56]

    Addressing function approximation error in actor-critic methods

    Scott Fujimoto, Herke Hoof, and David Meger. Addressing function approximation error in actor-critic methods. InProceedings of the 35th International Conference on Machine Learning (ICML), pages 1582–1591, 2018

    2018

  49. [57]

    Lillicrap, Jonathan J

    Timothy P . Lillicrap, Jonathan J. Hunt, Alexander Pritzel, Nicolas Heess, Tom Erez, Y uval Tassa, David Silver, and Daan Wierstra. Continuous con- trol with deep reinforcement learning.arXiv preprint arXiv:1509.02971, 2015

    2015 arXiv

  50. [58]

    Optuna: A next-generation hyperparameter optimiza- tion framework

    Takuya Akiba, Shotaro Sano, Toshihiko Y anase, Takeru Ohta, and Masanori Koyama. Optuna: A next-generation hyperparameter optimiza- tion framework. InProceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD ’19, page 2623–2631, New ...

    2019

  51. [59]

    Computing Resources – HAICORE

    Helmholtz AI. Computing Resources – HAICORE. https://www. helmholtz.ai/you-helmholtz-ai/computing-resources/, 2026

    2026

  52. [60]

    Lundberg and Su-In Lee

    Scott M. Lundberg and Su-In Lee. A unified approach to interpreting model predictions. InProceedings of the 31st International Conference on Neural Information Processing Systems (NIPS), pages 4765–4774, 2017

    2017

  53. [61]

    Craven and Jude W

    Mark W. Craven and Jude W. Shavlik. Extracting tree-structured repre- sentations of trained networks. InProceedings of the 8th International Conference on Neural Information Processing Systems (NIPS), pages 24– 30, 1996

    1996

  54. [62]

    A comparative study of deep reinforcement learning models: Dqn vs ppo vs a2c.arXiv preprint arXiv:2407.14151, 2024

    Neil De La Fuente and Daniel A Vidal Guerra. A comparative study of deep reinforcement learning models: Dqn vs ppo vs a2c.arXiv preprint arXiv:2407.14151, 2024

    2024

  55. [63]

    Artificial Intelligence Ac, 2024

    European Comission. Artificial Intelligence Ac, 2024

    2024

  56. [64]

    Department of Homeland Security

    U.S. Department of Homeland Security. Roles and Responsibilities Frame- work for Artificial Intelligence in Critical Infrastructure, 2024

    2024

  57. [65]

    Learning to run a power network chal- lenge: a retrospective analysis, 2021

    Antoine Marot, Benjamin Donnot, Gabriel Dulac-Arnold, Adrian Kelly, Aïdan O’Sullivan, Jan Viebahn, Mariette Awad, Isabelle Guyon, Patrick Panciatici, and Camilo Romero. Learning to run a power network chal- lenge: a retrospective analysis, 2021. 14 VOLUME 00, 0000 Preprint sub...

    2021

  58. [66]

    ERFAN TAJALLI-ARDEKANIreceived the B.Sc

    In 2021, he became the Head of Group for the research group IT Methods and Components for Energy Systems (IT4ES) with the Institute of Automation and Applied Informatics, KIT. ERFAN TAJALLI-ARDEKANIreceived the B.Sc. degree in industrial engineering and manage- ment and the M....

    2021

Pith tools

Reviewed June 28, 2026 · model on record in the stance chip above.