Article
Computer Science, Information Systems
Rebekka Olsson Omslandseter, Lei Jiao, B. John Oommen
Summary: This paper introduces a new approach called Hierarchical Discrete Pursuit Automaton (HDPA) which enhances the convergence capabilities of the traditional two-action Learning Automata (LA) methods by utilizing estimates to order the actions. The results show that HDPA is faster and more accurate than previous LA methods.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Information Systems
Divija Swetha Gadiraju, V. Lalitha, Vaneet Aggarwal
Summary: Blockchains have shown high levels of security and reliability in various applications. Prism is a new blockchain algorithm that achieves maximum throughput and minimal latency without compromising security. This study applies Deep Reinforcement Learning (DRL) to optimize the performance of Prism and proposes a DRL-based Prism Blockchain (DRLPB) scheme. Two widely used DRL algorithms, Dueling Deep Q Networks (DDQN) and Proximal Policy Optimization (PPO), are applied in DRLPB to compare their performance. The DRLPB scheme enhances the number of votes by up to 84% compared to Prism, while maintaining the security and latency performance guarantees.
IEEE TRANSACTIONS ON SERVICES COMPUTING
(2023)
Article
Energy & Fuels
Alejandro Ramirez Arango, Jose Aguilar, Maria D. R-Moreno
Summary: This study proposes Deep Reinforcement Learning approaches to solve the complex hydro-thermal economic dispatch problem, which can handle uncertainty and sequential decisions. The performance of these approaches is compared with a classic deterministic method, and the advantages of the proposed methods are highlighted.
SUSTAINABLE ENERGY GRIDS & NETWORKS
(2023)
Article
Computer Science, Hardware & Architecture
Abeer Z. Al-Marridi, Amr Mohamed, Aiman Erbad
Summary: Emerging technological innovation towards e-Health transition is a global priority for ensuring people's quality of life. The proposed Healthchain-RL framework combines Blockchain technology and artificial intelligence to optimize medical data exchange among heterogeneous healthcare organizations, addressing security, latency, and cost trade-offs effectively. By leveraging Deep Reinforcement Learning algorithms, the system achieves real-time adaptivity while maintaining maximum security and minimum latency and cost.
Article
Computer Science, Artificial Intelligence
Xin Li, Haojie Lei, Li Zhang, Mingzhong Wang
Summary: This paper explores interpretable Deep Reinforcement Learning (DRL) by representing policy using Differentiable Inductive Logic Programming (DILP). The research focuses on the optimization perspective of DILP-based policy learning and proposes using Mirror Descent for policy optimization. The theoretical and empirical studies verify the effectiveness of the proposed approach.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Energy & Fuels
Junshuang Zhou, Juncheng Zhou, Yue Xu, Bingchang Yan, Xinyue Yu
Summary: By altering the contribution of oxygen species, the three methane plasma oxidative pathways can be strengthened. The reaction of O(D-1) + CH4 -> OH + CH3 plays a key role in plasma promoted methane conversion, while the ground state of O leads to synthesis gas production through the reaction with CH3 and CH2. This discovery may provide insight into methane plasma oxidation and combustion chemistry.
Review
Mathematics
Bruno Gasperov, Stjepan Begusic, Petra Posedel Simovic, Zvonko Kostanjcar
Summary: Market making involves a market participant continuously posting limit orders to provide liquidity and profit, optimal market making requires dynamic adjustment of bid and ask prices to maximize risk-adjusted returns, reinforcement learning techniques outperform standard strategies in terms of performance.
Article
Medicine, General & Internal
Juan Jovel, Russell Greiner
Summary: Machine learning approaches, including supervised, unsupervised, and reinforcement learning, are increasingly popular in biomedical research for various applications such as drug design, patient stratification, and medical image analysis. These methods have the potential to greatly impact basic and applied research programs in biomedicine.
FRONTIERS IN MEDICINE
(2021)
Article
Thermodynamics
Dan Hou, Jiayu Huang, Yanyu Wang
Summary: Building performance optimization (BPO) is a common method in energy-efficient building design. This study introduces three categories of constrained optimization approaches and makes improvements to better match practical problems. Results show that biased bi-objective optimization achieves the best performance on average, while reinforcement learning guided optimization is affected by the capacity of the RL model. Penalty functions are not robust and efficient enough in solving practical BPO problems. Further investigation reveals that preserving good infeasible solutions is more effective than simply improving the proportion of feasible solutions.
Article
Chemistry, Physical
Rodrigo F. B. de Souza, Daniel Z. Florio, Ermete Antolini, Almir O. Neto
Summary: This work reviews the utilization of fuel cells for the simultaneous generation of electricity and valuable chemicals through the partial oxidation of methane.
Article
Engineering, Electrical & Electronic
Niloy Saha, Mohammad Zangooei, Morteza Golkarifard, Raouf Boutaba
Summary: Network slicing in 5G and beyond networks customizes the network for each application or service by chaining virtualized network functions (VNFs) according to service requirements. The increased flexibility of network slicing comes with management complexity that cannot be solved by traditional solutions, and therefore requires minimizing human intervention through the use of artificial intelligence techniques. This article surveys various deep reinforcement learning (DRL)-based approaches to slice scaling and placement, highlighting their benefits and addressing key challenges and open issues.
IEEE COMMUNICATIONS MAGAZINE
(2023)
Article
Computer Science, Artificial Intelligence
Abida Sharif, Jian Ping Li, Muhammad Asim Saleem, Gunasekaran Manogran, Seifedine Kadry, Abdul Basit, Muhammad Attique Khan
Summary: The Internet of Vehicles (IoV) connects vehicles to the Internet to transfer information, and network clustering strategies are proposed to solve traffic management challenges in IoV networks. Reinforcement learning is used to learn optimal policies, and an experience-driven approach based on deep reinforcement learning is proposed for efficiently selecting cluster heads in managing network resources in the IoV environment.
JOURNAL OF INTELLIGENT MANUFACTURING
(2021)
Article
Chemistry, Multidisciplinary
Bo Wu, Tiejun Lin, Min Huang, Shenggang Li, Ji Li, Xing Yu, Ruoou Yang, Fanfei Sun, Zheng Jiang, Yuhan Sun, Liangshu Zhong
Summary: A bimetallic catalyst PdCu/Z-5 was developed for the selective oxidation of methane to oxygenates at low temperature, achieving high yield and selectivity through the synergistic effect of PdO nanoparticles and Cu single atoms.
ANGEWANDTE CHEMIE-INTERNATIONAL EDITION
(2022)
Article
Automation & Control Systems
Mridul Agarwal, Vaneet Aggarwal
Summary: The paper discusses the problem of optimizing a non-linear function of the long term average rewards and proposes model-based and model-free algorithms to solve it. The proposed algorithms are shown to outperform conventional RL approaches in fairness scheduling and queueing system scheduling.
JOURNAL OF MACHINE LEARNING RESEARCH
(2023)
Article
Economics
Zengxiang Lei, Satish V. Ukkusuri
Summary: This study proposes a reinforcement learning-based approach for the dynamic pricing problem in ride-hailing systems. By translating the problem into a Markov Decision Process, the existence of a deterministic stationary optimal policy is proven. Using the offline learning algorithm TD3, the optimal pricing policy is learned from historical data and applied to the next time slot. Extensive numerical experiments demonstrate the effectiveness of the proposed algorithm in finding the optimal pricing policy and improving platform profit and service efficiency in both small and large networks.
TRANSPORTATION RESEARCH PART B-METHODOLOGICAL
(2023)
Article
Chemistry, Physical
Regina Palkovits, Stefan Palkovits
Article
Chemistry, Multidisciplinary
Bruna Ferreira Gomes, Fabian Joschka Holzhaeuser, Carlos Manuel Silva Lobo, Pollyana Ferreira da Silva, Ernesto Danieli, Marcelo Carmo, Luiz Alberto Colnago, Stefan Palkovits, Regina Palkovits, Bernhard Bluemich
ACS SUSTAINABLE CHEMISTRY & ENGINEERING
(2019)
Article
Chemistry, Physical
Stefan Palkovits
Article
Multidisciplinary Sciences
Niklas Julian Lentelink, Stefan Palkovits
ADVANCED THEORY AND SIMULATIONS
(2020)
Article
Chemistry, Multidisciplinary
Nils Kurig, Jerome Meyers, F. Joschka Holzhaeuser, Stefan Palkovits, Regina Palkovits
Summary: This article discusses the concept of transforming Kolbe chemistry into flow processes, achieving continuous production through recirculation and single-pass systems. Non-Kolbe electrolysis was found to be better suited for continuous production, and an example was presented using different flow cells to convert chemicals.
ACS SUSTAINABLE CHEMISTRY & ENGINEERING
(2021)
Article
Chemistry, Physical
Christoph Wulf, Matthias Beller, Thomas Boenisch, Olaf Deutschmann, Schirin Hanf, Norbert Kockmann, Ralph Kraehnert, Mehtap Oezaslan, Stefan Palkovits, Sonja Schimmler, Stephan A. Schunk, Kurt Wagemann, David Linke
Summary: Modern research methods generate significant scientific data, but access to high-quality data remains limited in many fields. Implementing the FAIR data concept in the catalysis community could dramatically improve this situation. The German NFDI initiative aims to establish a comprehensive research data infrastructure across all scientific disciplines.
Article
Electrochemistry
Maike Berger, Ioana M. Popa, Leila Negahdar, Stefan Palkovits, Bastian Kaufmann, Moritz Pilaski, Harry Hoster, Regina Palkovits
Summary: Nickel-iron layered double hydroxides (NiFe LDH) are efficient electrocatalysts for alkaline oxygen evolution reaction (OER). This study investigated the kinetic behavior of NiFe LDH catalysts with different intercalated anions using steady-state Tafel plots, impedance measurements, and reaction order plots. The results showed that intercalated anions affect the rate-determining steps and have a significant impact on the OER activity.
Article
Chemistry, Multidisciplinary
Jerome Meyers, Joel B. Mensah, F. Joschka Holzhaeuser, Ahmad Omari, Christian C. Blesken, Till Tiso, Stefan Palkovits, Lars M. Blank, Stefan Pischinger, Regina Palkovits
ENERGY & ENVIRONMENTAL SCIENCE
(2019)