Article
Computer Science, Information Systems
Csaba Veres
Summary: Natural Language Processing (NLP) is a prominent application area in the field of Artificial Intelligence. Transfer learning has greatly enhanced the performance of deep learning neural networks trained on language modeling tasks. These language models, when trained with software code data, have shown remarkable capabilities in generating functional computer code from natural language specifications. This challenges the claim that eliminative neural models can replace symbolic abstractions like generative phrase structure grammars, as programming language syntax is determined by phrase structure grammars. Furthermore, the success of neural models in tasks involving symbolic systems proves that language and cognitive systems are indeed symbolic. As a result, the term "language model" is deemed misleading and the adoption of the term "corpus model" is proposed.
Article
Multidisciplinary Sciences
V. Balaji, Fleur Couvreux, Julie Deshayes, Jacques Gautrais, Frederic Hourdinf, Catherine Rio
Summary: Traditional general circulation models (GCMs) have been a main tool in climate research, but they have limitations. Future models can address these limitations through higher resolution and machine learning techniques. Calibration is key to understanding the inner workings of complex systems. GCMs will continue to play a central role in future climate research.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)
Review
Plant Sciences
Ginga Shimakawa
Summary: This review summarizes the molecular mechanisms of electron transport in cyanobacterial thylakoid membranes to enhance our understanding of the complexity of cyanobacterial electron transport. Cyanobacteria are the simplest oxygenic phototrophs structurally, but the regulation of their photosynthesis is challenging due to the shared thylakoid membranes and cytosolic space for photosynthetic and respiratory processes. The review aims to summarize the molecular mechanisms and in vivo activities of electron transport in cyanobacterial thylakoid membranes based on the latest research progress in photosynthesis.
JOURNAL OF EXPERIMENTAL BOTANY
(2023)
Article
Biochemical Research Methods
Srilakshmi Pattabiraman, Tandy Warnow
Summary: Profile Hidden Markov Models (HMMs) are graphical models that generate finite length sequences from a distribution, widely used in bioinformatics. The construction of profile HMMs is a statistical estimation problem, and it is unknown whether they are statistically identifiable.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2021)
Article
Physics, Multidisciplinary
Feng Liu, Zhenhao Fan, Zhipeng Sun, Xuzong Chen, Dingping Li
Summary: The systematic non-perturbative method developed based on Dyson-Schwinger theory and the derivable theory for Ising model at broken phase allows for obtaining critical temperature and spin spin correlation beyond mean field theory. The gapless spectrum of Green function obtained from this method at critical point results in divergent susceptibility at T-c. The critical temperature of Ising model obtained from this method is fairly good compared to other non-cluster methods and can be extended to more complex spin models with continuous symmetry.
FRONTIERS OF PHYSICS
(2021)
Article
History & Philosophy Of Science
Caspar Jacobs
Summary: This paper examines a counter-argument to the issue of underdetermination in physics, which claims that the representational capacities of models are irrelevant and the issue concerns possible worlds themselves. The paper refutes two versions of this objection, one regarding the formalism of a theory and the other concerning the notion of possibility.
Article
Multidisciplinary Sciences
Rohan Rao, Jing Hu, Po-Heng Lee
Summary: Direct interspecies electron transfer (DIET) is an efficient metabolism between symbiotically interacting organisms, which utilizes conductive materials to transfer electrons from electron donating organisms to electron accepting organisms. Conductive materials become negatively charged in DIET processes and electron tunnelling becomes a significant electron transfer mechanism.
SCIENTIFIC REPORTS
(2022)
Article
Mathematics
Erfan Khaniki
Summary: This paper focuses on the local properties of Kripke models of Heyting Arithmetic. By introducing two new Kripke model construction methods, the question of whether every Kripke model is locally PA is answered negatively, and several related results and applications are obtained.
ADVANCES IN MATHEMATICS
(2022)
Article
Cell Biology
Yusuke Fujiwara, Chenyang Ding, Yohei Sanada, Dilimulati Yimiti, Masakazu Ishikawa, Tomoyuki Nakasa, Naosuke Kamei, Kazunori Imaizumi, Martin K. Lotz, Takayuki Akimoto, Shigeru Miyaki, Nobuo Adachi
Summary: This study investigated the role of miR-23a/b clusters in osteoarthritis (OA). Through experiments on genetically modified mice, it was found that miR-23a/b clusters do not significantly contribute to the pathogenesis of OA.
FRONTIERS IN CELL AND DEVELOPMENTAL BIOLOGY
(2023)
Article
Genetics & Heredity
Yichen Si, Brett Vanderwerff, Sebastian Zollner
Summary: Genotype imputation is crucial in human genetic studies, but current algorithms may have model misspecifications that limit the accurate imputation of rare variants. The probability of falsely imputed minor alleles decreases with reference sample size, but the proportion of erroneously imputed minor alleles depends on the allele count in the reference sample.
Article
Agronomy
Kritika Kothari, Rafael Battisti, Kenneth J. Boote, Sotirios Archontoulis, Adriana Confalone, Julie Constantin, Santiago Cuadra, Philippe Debaeke, Babacar Faye, Brian Grant, Gerrit Hoogenboom, Qi Jing, Michael van der Laan, Fernando Antonio Macena da Silva, Fabio R. Marin, Alireza Nehbandani, Claas Nendel, Larry C. Purcell, Budong Qian, Alex C. Ruane, Celine Schoving, Evandro H. F. M. Silva, Ward Smith, Afshin Soltani, Amit Srivastava, Nilson A. Vieira, Stacey Slone, Montserrat Salmeron
Summary: Accurately estimating crop yield under climate change scenarios is crucial for meeting food demands. However, there is a lack of comprehensive assessment for soybean, the main source of protein, in terms of climate change impact. In this study, ten models were used to simulate soybean yield under varying temperature and CO2 concentration. The results showed significant variability among models, highlighting the need for further development and experimentation to reduce uncertainty.
EUROPEAN JOURNAL OF AGRONOMY
(2022)
Article
Ecology
Michael N. Weiss, Daniel W. Franks, Lauren J. N. Brent, Samuel Ellis, Matthew J. Silk, Darren P. Croft
Summary: Datastream permutations do not typically represent the null hypothesis of interest for researchers interfacing animal social network analysis with regression modeling, and using this methodology may lead to potential pitfalls. Simulations demonstrate that using datastream permutations to indicate relationships between network structure and covariates can result in extremely high type I error rates.
METHODS IN ECOLOGY AND EVOLUTION
(2021)
Article
Environmental Sciences
Alexander Koch, Wannes Hubau, Simon L. Lewis
Summary: Tropical forests have been a significant carbon sink in the past few decades, but recent research shows that this sink is saturated and declining in the long term. Climate models and statistical models provide different projections on the future carbon sink, with potential impacts of climate change on tropical forest ecosystems. The discrepancies between model projections and observational data highlight the importance of accurately representing vegetation mortality in future climate projections.
Review
Health Care Sciences & Services
Uwe M. Pommerich, Peter W. Stubbs, Peter Preben Eggertsen, Jesper Fabricius, Jorgen Feldbaek Nielsen
Summary: This study aimed to identify and summarize validated multivariable prognostic models for assessing functional independence in patients with acquired brain injury (ABI) at discharge from post-acute rehabilitation. Three articles met the inclusion criteria, with a total of six internally validated and two externally validated prognostic models. However, these models were found to have limitations in terms of reliability and accuracy, indicating that they are not yet suitable for clinical use.
JOURNAL OF CLINICAL EPIDEMIOLOGY
(2023)
Article
Mechanics
Junghaeng Lee, Taehoon Kim, Hyunkyu Jang, Mikyung Kwon, Kwang Soo Cho
Summary: In this study, various rheological models such as Palierne, Gramespacher-Meissner, and Doi-Ohta models were compared based on irreversible thermodynamics for the immiscible blend of fluids. It was found that the dynamic modulus of all the models could be expressed as the sum of polymer modulus and interfacial modulus. The interfacial modulus of the three models was shown to be a single-mode Maxwell model. Based on the comparison with experimental data, the thermodynamic model with the linearized Doi-Ohta equation was concluded to be more robust than the other models.
JOURNAL OF NON-NEWTONIAN FLUID MECHANICS
(2023)