Article
Computer Science, Information Systems
Haobin Shi, Jingchen Li, Shicong Chen, Kao-Shing Hwang
Summary: Inverse reinforcement learning is used in deep reinforcement learning systems to tackle tasks that are difficult to design with manual reward functions. This study proposes a behavior fusion method based on adversarial IRL to overcome the issue of varying reward functions in complex tasks. The method decomposes tasks into subtasks and uses multiple discriminators to improve the adversarial IRL model. Experimental results show that the proposed method can learn advanced policies and ensure a stable training process.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Yi-Feng Zhang, Fan-Ming Luo, Yang Yu
Summary: Imitation learning aims to recover expert policies from limited demonstration data. Generative Adversarial Imitation Learning (GAIL) has shown great potential in using the generative adversarial learning framework for imitation learning. However, GAIL and its variants are highly sensitive to hyperparameters and difficult to converge well. One key issue is the faster learning speed of the supervised learning discriminator compared to the reinforcement learning generator, resulting in vanishing generator gradients.
Article
Chemistry, Multidisciplinary
Yuxia Tang, Jiulou Zhang, Doudou He, Wenfang Miao, Wei Liu, Yang Li, Guangming Lu, Feiyun Wu, Shouju Wang
Summary: The study introduces a Generative Adversarial Network for Distribution Analysis (GANDA) to describe and generate intratumoral quantum dots distribution, offering a new approach to investigate NPs distribution and guide nanomedicine optimization.
JOURNAL OF CONTROLLED RELEASE
(2021)
Review
Computer Science, Information Systems
Guillermo Iglesias, Edgar Talavera, Alberto Diaz-Alvarez
Summary: In recent years, deep learning has been revolutionized by the significant impact of Generative Adversarial Networks (GANs), which provide a unique architecture and generate incredible results. Due to the continuous development and wide range of applications, keeping up with the latest research in GANs becomes challenging. This survey aims to provide an overview of GANs, including the latest architectures, optimizations, validation metrics, and application areas, with the goal of guiding future researchers in achieving better results.
COMPUTER SCIENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Rui Li, Zhengbo Zou
Summary: The development and deployment of robots on construction sites are crucial for the industrialization of construction known as Construction 4.0. However, their utilization on-site is limited due to the need for expert remote control and adaptability challenges. Reinforcement learning, particularly inverse reinforcement learning methods like GAIL, can effectively enhance robot performance in tackling complex construction tasks without explicitly defined reward functions. The proposed VR-GAIL model achieved a 4.5% higher success rate compared to the baseline RL model PPO, indicating its effectiveness and potential in improving construction task efficiency. Rating: 8/10
ADVANCED ENGINEERING INFORMATICS
(2023)
Article
Robotics
Huixin Zhan, Feng Tao, Yongcan Cao
Summary: In this study, a new reinforcement learning approach based on human preferences is proposed, utilizing a generative adversarial network to reduce the need for human queries and designing a maximum entropy based reinforcement learning algorithm. The method shows significant reduction in human time without sacrificing performance when applied to complex robotic tasks.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Information Systems
Qi Wang, Yongsheng Hao, Jiawei Zhang
Summary: This paper proposes GIRL (Generative Inverse Reinforcement Learning), a method that combines generative adversarial networks (GANs) and deep reinforcement learning (DRL) to solve combinatorial optimization problems. GIRL can learn heuristics without explicit extrinsic rewards and improves existing methods by enhancing generalization capabilities, improving accuracy, and addressing sparse reward problems.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2023)
Review
Materials Science, Multidisciplinary
Addis S. Fuhr, Bobby G. Sumpter
Summary: Machine learning and artificial intelligence methods are making a significant impact on chemistry and condensed matter physics. Deep learning methods have shown new capabilities in high-throughput virtual screening and inverse design of materials. Deep generative models (GMs) encode material structure and properties into a latent space to generate new materials. Applying GMs to inorganic materials has been challenging, but recent innovations have enabled the acceleration of inorganic materials discovery.
FRONTIERS IN MATERIALS
(2022)
Article
Computer Science, Artificial Intelligence
Lionel Blonde, Pablo Strasser, Alexandros Kalousis
Summary: This paper investigates off-policy generative adversarial imitation learning and emphasizes the importance of making the learned reward function locally Lipschitz-continuous for the method to perform well. Through theoretical and empirical research, several theoretical results involving the local Lipschitzness of the state-value function are provided, and the positive effect of satisfying the Lipschitzness constraint on imitation performance is demonstrated. Additionally, a generic pessimistic reward preconditioning add-on is proposed, which enhances the robustness of the base method and is supported by several theoretical guarantees.
Article
Computer Science, Information Systems
Tuyen Pham Le, Cheolkyun Rho, Yelin Min, Sungreong Lee, Daewoo Choi
Summary: This paper proposes a novel framework based on reinforcement learning and generative adversarial networks to address the issues in financial data, resulting in a significant reduction in financial risk.
Article
Computer Science, Artificial Intelligence
Yueyue Hu, Shiliang Sun
Summary: This paper investigates the adversarial robustness of RL agents and proposes a novel defense framework RL-VAEGAN based on the idea of style transfer. The framework effectively defends against state-of-the-art methods in white-box and black-box scenarios with diverse magnitudes of perturbations.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Yirui Zhou, Mengxiao Lu, Xiaowei Liu, Zhengping Che, Zhiyuan Xu, Jian Tang, Yangchun Zhang, Yan Peng, Yaxin Peng
Summary: Generative adversarial imitation learning (GAIL) treats imitation learning (IL) as a distribution matching problem between expert policy and learned policy. This paper focuses on the generalization and computational properties of policy classes. It is proven that GAIL can ensure generalization when policies are well controlled. By incorporating distributional reinforcement learning (RL) into GAIL, the greedy distributional soft gradient (GDSG) algorithm is proposed to solve GAIL. GDSG has advantages including alleviating Q-value overestimation problem and improving policy performance through sufficient exploration, as well as attaining a sublinear convergence rate to a stationary solution. Comprehensive experimental verification in MuJoCo environments demonstrates that GDSG outperforms previous GAIL variants in mimicking expert demonstrations.
Article
Computer Science, Artificial Intelligence
Zhihao Cheng, Li Shen, Miaoxi Zhu, Jiaxian Guo, Meng Fang, Liu Liu, Bo Du, Dacheng Tao
Summary: This paper proposes an algorithm that can adaptively learn safe policies from a single expert dataset under diverse safety constraints. It introduces the use of a Lagrange multiplier to balance imitation and safety performance, and employs a two-stage optimization framework to solve the problem. Experimental results demonstrate the effectiveness of the approach.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Hady Pranoto, Yaya Heryadi, Harco Leslie Hendric Spits Warnars, Widodo Budiharto
Summary: Cross aging affects face recognition ability, and using synthetic face images can improve performance. A new optimized variant is proposed for generating synthetic face images at specific age groups, with improvements in accuracy and training time. Evaluation results demonstrate better accuracy in various tasks.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Biochemical Research Methods
Fengqing Lu, Mufei Li, Xiaoping Min, Chunyan Li, Xiangxiang Zeng
Summary: This study introduces a computational framework called DLGN for generating bioactive molecules towards two specific targets. DLGN utilizes adversarial training and reinforcement learning to explore chemical spaces and encourage the generation of molecules that belong to the intersection of two bioactive compound distributions. The proposed model shows promise in generating novel compounds with high similarity to multiple bioactive datasets.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Geriatrics & Gerontology
Mikhail V. V. Shaposhnikov, Anastasia A. A. Gorbunova, Nadezhda V. V. Zemskaya, Natalia S. S. Ulyasheva, Natalya R. R. Pakshina, Daria V. V. Yakovleva, Alexey Moskalev
Summary: This study investigated the geroprotective effects of individual and simultaneous overexpression of genes encoding key enzymes of H2S biosynthesis on a Drosophila melanogaster model. Simultaneous overexpression of CBS and CSE showed additive effects in males and synergistic effects in females on median lifespan. Individual overexpression of CBS increased thermotolerance and decreased transcription levels of stress-responsive transcription factors HIF1 and Hsf, while individual overexpression of CSE increased paraquat resistance. Simultaneous overexpression of both genes increased resistance to hyperthermia in old females or paraquat in old males. The findings suggest a sex-specific epistatic interaction between CBS and CSE overexpression effects on longevity and stress resistance.
Letter
Gastroenterology & Hepatology
Daniil Rafaelevich Markaryan, Aleksandr Maksimovich Lukianov, Tatiana Nikolaevna Garmanova, Ekaterina Aleksandrovna Kazachenko, Aleksey Igorevich Moskalev, Bruno Roche
COLORECTAL DISEASE
(2023)
Article
Chemistry, Medicinal
Yan A. Ivanenkov, Daniil Polykovskiy, Dmitry Bezrukov, Bogdan Zagribelnyy, Vladimir Aladinskiy, Petrina Kamya, Alex Aliper, Feng Ren, Alex Zhavoronkov
Summary: Chemistry42 is a software platform that combines Artificial Intelligence techniques with computational and medicinal chemistry methodologies for de novo small molecule design and optimization. It efficiently generates novel molecular structures with optimized properties, which are validated in both in vitro and in vivo studies. Chemistry42 is part of Insilico Medicine's Pharma.ai drug discovery suite, along with PandaOmics for target discovery and multiomics data analysis, and inClinico for data-driven forecast of a clinical trial's probability of success.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2023)
Article
Chemistry, Multidisciplinary
Feng Ren, Xiao Ding, Min Zheng, Mikhail Korzinkin, Xin Cai, Wei Zhu, Alexey Mantsyzov, Alex Aliper, Vladimir Aladinskiy, Zhongying Cao, Shanshan Kong, Xi Long, Bonnie Hei Man Liu, Yingtao Liu, Vladimir Naumov, Anastasia Shneyderman, Ivan V. Ozerov, Ju Wang, Frank W. Pun, Daniil A. Polykovskiy, Chong Sun, Michael Levitt, Alan Aspuru-Guzik, Alex Zhavoronkov
Summary: The application of artificial intelligence (AI) in drug discovery has revolutionized the field. The AlphaFold program's ability to predict protein structures for the entire human genome is considered a remarkable breakthrough. By applying AlphaFold, we successfully identified a novel hit molecule for a novel target without experimental structure.
Article
Cell Biology
Elena Yushkova, Alexey Moskalev
Summary: Transposable elements (TEs) play a crucial role in eukaryotic genomes, impacting aging, carcinogenesis, and other age-related diseases. This review explores the fundamental properties of TEs and their complex interactions with cellular processes, highlighting their diverse effects on genetics and epigenetics. The review discusses TEs' interactions with recombination, replication, repair, and chromosomal regulation, their ability to balance activity and repression, their involvement in gene creation and RNA expression, and their role in DNA damage and regulatory networks. The review also evaluates the contribution of derepressed TEs to age-related effects in individual cells and tissues. Conflicting information about TE activity under stress and theories related to aging mechanisms are addressed. Furthermore, the review examines the specific impact of TEs on aging processes in germline and soma, as well as the regulation of TEs in cells. Recent findings on somatic mutations in human and animal tissues are discussed, focusing on their potential functional consequences. Additionally, the review explores the correlation between somatic TE activation and age-related changes in heterochromatin maintenance and longevity regulation. Notably, the review also explores the differences between transposon- and retrotransposon-mediated structural genome changes and their association with aging and age-related pathologies. Finally, based on published data, the review proposes a hypothesis regarding the influence of species-specific features of TE number, composition, and distribution on aging dynamics in different animal genomes.
AGEING RESEARCH REVIEWS
(2023)
Review
Nutrition & Dietetics
Anastasiia V. Badaeva, Alexey B. Danilov, Paul Clayton, Alexey A. Moskalev, Alexander V. Karasev, Andrey F. Tarasevich, Yulia D. Vorobyeva, Viacheslav N. Novikov
Summary: Neuronutrition is a part of nutritional neuroscience that studies the effects of different dietary components on behavior and cognition. It also encompasses the use of various nutrients and diets to prevent and treat neurological disorders. This narrative review explores the current understanding of neuronutrition as a key concept for brain health and its potential application in the prevention and treatment of various disorders.
Article
Biochemistry & Molecular Biology
Fedor I. I. Isaev, Arsenii R. R. Sadykov, Alexey Moskalev
Summary: Based on blood parameters, we investigated the influence of Kivach Clinic's special medical spa program on the biological age of patients. The results showed that the spa treatment has the potential to reduce biological age.
Review
Pharmacology & Pharmacy
Frank W. Pun, Ivan V. Ozerov, Alex Zhavoronkov
Summary: Disease modeling and target identification are crucial in drug discovery, and artificial intelligence is increasingly being used in this field. This article reviews recent advances in AI-driven target discovery and discusses the importance of striking a balance between novelty and confidence in target selection. It also highlights the validation of AI-identified targets through experiments and the potential pathways for further development.
TRENDS IN PHARMACOLOGICAL SCIENCES
(2023)
Article
Chemistry, Medicinal
Yan Ivanenkov, Bogdan Zagribelnyy, Alex Malyshev, Sergei Evteev, Victor Terentiev, Petrina Kamya, Dmitry Bezrukov, Alex Aliper, Feng Ren, Alex Zhavoronkov
Summary: This article provides a comprehensive overview of the most recent research outcomes on artificial intelligence-generated molecular structures from the perspective of medicinal chemists. The focus is on studies that involve synthesis, in vitro validation, and analysis of the relevance and novelty of these structures in modern medicinal chemistry. The authors believe that this review will be valuable to the medicinal chemistry and AI-driven drug design communities as a comprehensive approach for evaluating different research outcomes in AIDD.
ACS MEDICINAL CHEMISTRY LETTERS
(2023)
Article
Biotechnology & Applied Microbiology
A. P. Karmanov, L. S. Kocheva, O. V. Raskosha, A. A. Moskalev
Summary: This paper presents studies on the chemical and topological structure of lignin macromolecules from various herbaceous plants. The composition and properties of the biopolymers were determined using 13C-NMR spectroscopy, electron paramagnetic resonance, FTIR spectroscopy, and Py-GC/MS spectrometry. The results show that lignin from Rhodiola rosea belongs to the class of star-shaped polymers, and it has geroprotective properties and improves the cognitive abilities of model animals.
BIOCATALYSIS AND AGRICULTURAL BIOTECHNOLOGY
(2023)
Article
Biochemistry & Molecular Biology
Daria V. Mikhailova, Oksana G. Shevchenko, Denis A. Golubev, Elena Y. Platonova, Nadezhda V. Zemskaya, Olesya Yu. Shoeva, Elena I. Gordeeva, Sergey A. Patov, Mikhail V. Shaposhnikov, Elena K. Khlestkina, Alexey Moskalev
Summary: This study examines the antioxidant and geroprotective properties of wheat bran extracts with high anthocyanin levels. The results show that the extracts with anthocyanins have higher radical scavenging and membrane protective activities, and can extend the lifespan of Drosophila. However, other metabolites in wheat bran may also contribute to its antioxidant and geroprotective potential.
Review
Geriatrics & Gerontology
Nicola Marino, Guido Putignano, Simone Cappilli, Emmanuele Chersoni, Antonella Santuccione, Giuliana Calabrese, Evelyne Bischof, Quentin Vanhaelen, Alex Zhavoronkov, Bryan Scarano, Alessandro D. Mazzotta, Enrico Santus
Summary: In the past, technology was mainly used to store information about protein and molecular structures, but now artificial intelligence can learn from existing data to predict and model properties and interactions, revealing important knowledge about complex biological processes. Modern technologies can also utilize more information to understand the interactions between the human body and the external environment, particularly the role of external factors in aging. As computational systems biology improves and new biomarkers of aging are developed, artificial intelligence promises to be a major ally in aging research.
FRONTIERS IN AGING
(2023)
Article
Cell Biology
Frank W. Pun, Geoffrey Ho Duen Leung, Hoi Wing Leung, Bonnie Hei Man Liu, Xi Long, Ivan Ozerov, Ju Wang, Feng Ren, Alexander Aliper, Evgeny Izumchenko, Alexey Moskalev, Joao Pedro de Magalhaes, Alex Zhavoronkov
Summary: Aging biology is a promising field, and this study proposes a list of novel aging-associated targets and classical targets for drug discovery and repurposing. Most of the top targets play a role in inflammation and extracellular matrix stiffness, highlighting the relevance of these processes in aging and age-related diseases. The PandaOmics platform demonstrates its application in target discovery across multiple disease areas.
Article
Cell Biology
Irina Strazhesko, Olga Tkacheva, Daria Kashtanova, Mikhail Ivanov, Vladislav Kljashtorny, Antonina Esakova, Maria Karnaushkina, Cassandra Guillemette, Amber Hewett, Veronique Legault, Lilit Maytesian, Maria Litvinova, Alan Cohen, Alexey Moskalev
Summary: Old age is a crucial risk factor for severe COVID-19, and physiological health status and biological age are important factors that affect COVID-19 severity and mortality. Certain physiological indicators can predict the deterioration and death risk of COVID-19. These findings have significant implications for the management of COVID-19 patients.
Article
Cell Biology
Esther Meron, Maria Thaysen, Suzanne Angeli, Adam Antebi, Nir Barzilai, Joseph A. Baur, Simon Bekker-Jensen, Maria Birkisdottir, Evelyne Bischof, Jens Bruening, Anne Brunet, Abigail Buchwalter, Filipe Cabreiro, Shiqing Cai, Brian H. Chen, Maria Ermolaeva, Collin Y. Ewald, Luigi Ferrucci, Maria Carolina Florian, Kristen Fortney, Adam Freund, Anastasia Georgievskaya, Vadim N. Gladyshev, David Glass, Tyler Golato, Vera Gorbunova, Jan Hoejimakers, Riekelt H. Houtkooper, Sibylle Jager, Frank Jaksch, Georges Janssens, Martin Borch Jensen, Matt Kaeberlein, Gerard Karsenty, Peter de Keizer, Brian Kennedy, James L. Kirkland, Michael Kjaer, Guido Kroemer, Kai-Fu Lee, Jean-Marc Lemaitre, David Liaskos, Valter D. Longo, Yu-Xuan Lu, Michael R. MacArthur, Andrea B. Maier, Christina Manakanatas, Sarah J. Mitchell, Alexey Moskalev, Laura Niedernhofer, Ivan Ozerov, Linda Partridge, Emmanuelle Passegue, Michael A. Petr, James Peyer, Dina Radenkovic, Thomas A. Rando, Suresh Rattan, Christian G. Riedel, Lenhard Rudolph, Ruixue Ai, Manuel Serrano, Bjoern Schumacher, David A. Sinclair, Ryan Smith, Yousin Suh, Pam Taub, Alexandre Trapp, Anne-Ulrike Trendelenburg, Dario Riccardo Valenzano, Kris Verburgh, Eric Verdin, Jan Vijg, Rudi G. J. Westendorp, Alessandra Zonari, Daniela Bakula, Alex Zhavoronkov, Morten Scheibye-Knudsen
Summary: Aging is the biggest risk factor for most chronic diseases, attracting attention from industry and investors. This year's ARDD meeting featured presentations from 75 speakers and included a longevity workshop to discuss aging mechanisms and potential modifications.