A5100605566- Geology and Paleoclimatology Research
- Geochemistry and Elemental Analysis
- Geological and Geochemical Analysis
- Paleontology and Stratigraphy of Fossils
- Geological and Geophysical Studies
- Hydrocarbon exploration and reservoir analysis
- Geological formations and processes
- Geochemistry and Geologic Mapping
- Methane Hydrates and Related Phenomena
- Groundwater and Isotope Geochemistry
- Geological Studies and Exploration
- Evolution and Paleontology Studies
- Domain Adaptation and Few-Shot Learning
- earthquake and tectonic studies
- CO2 Sequestration and Geologic Interactions
- Pleistocene-Era Hominins and Archaeology
- Geomagnetism and Paleomagnetism Studies
- Aeolian processes and effects
- Neural Networks and Applications
- Isotope Analysis in Ecology
- Advanced Neural Network Applications
- Remote Sensing and Land Use
- Soil Carbon and Nitrogen Dynamics
- Geochemistry and Geochronology of Asian Mineral Deposits
- Heavy metals in environment
Nanjing Forestry University
2025
Zhejiang University
2023-2025
Institute of Tibetan Plateau Research
2015-2024
Chinese Academy of Sciences
2015-2024
National University of Malaysia
2024
Yanshan University
2024
University of Chinese Academy of Sciences
2014-2024
Institute of Coal Chemistry
2023-2024
University of Colorado Boulder
2023
Tianjin University of Science and Technology
2023
Abstract Plate-tectonic processes have long been thought to be the major cause of Cenozoic global carbon cycle, and cooling by uplift Tibetan Plateau through enhancing silicate weathering organic burial and/or obducted ophiolites during closure Neo-Tethys Ocean. However, imbalance resulting from accelerated CO2 consumption a relatively stable input volcanic degassing should depleted atmospheric within few million years; therefore, negative feedback mechanism must stabilized cycle. Here, we...
Convolution neural networks (CNNs) and Transformers have their own advantages both been widely used for dense prediction in multi-task learning (MTL). Most of the current studies on MTL solely rely CNN or Transformer. In this work, we present a novel model by combining merits deformable query-based Transformer prediction. Our method, named DeMT, is based simple effective encoder-decoder architecture (i.e., mixer encoder task-aware transformer decoder). First, contains two types operators:...
Transition-metal dichalcogenides (TMDs) have demonstrated a wide range of novel photonic, optoelectronic, and correlated electron phenomena for more than decade. However, the coherent dynamics their excitons, including possibly long dephasing times sensitivity to spatial heterogeneities, are still poorly understood. Here we implement adiabatic plasmonic nanofocused four-wave mixing (FWM) image in monolayer WSe2. We observe nanoscale heterogeneities at room temperature with ranging from T2 ≲...
Abstract The uplift of the Tibetan Plateau (TP) during late Cenozoic is thought to be one crucial factors controlling Asian climate. However, complex interaction between tectonics and climate change still unclear. Here we present first record clay mineralogy elemental geochemistry covering ~12.7–4.8 Ma in a fluvial‐lacustrine sequence Xining Basin. Geochemical provenance proxies (Th/Sc, Zr/Sc, Cr/Zr) <2‐μm fraction show significant at ~8.8 Ma. Silicate‐based weathering indexes (CIA, CIW,...
Research Article| February 26, 2019 Miocene fire intensification linked to continuous aridification on the Tibetan Plateau Yunfa Miao; Miao * 1Key Laboratory of Desert and Desertification, Northwest Institute Eco-Environment Resources, Chinese Academy Sciences, Lanzhou 730000, China2CAS Center for Excellence in Earth Sciences Key Continental Collision Uplift, Research, Beijing 100101, China *E-mails: miaoyunfa@lzb.ac.cn; wufuli@itpcas.ac.cn Search other works by this author on: GSW Google...
Few-shot class-incremental learning (FSCIL) has been a challenging problem as only few training samples are accessible for each novel class in the new sessions. Finetuning backbone or adjusting classifier prototypes trained prior sessions would inevitably cause misalignment between feature and of old classes, which explains well-known catastrophic forgetting problem. In this paper, we deal with dilemma FSCIL inspired by recently discovered phenomenon named neural collapse, reveals that...
Rare earth (RE)-based complexes, due to their unique electronic structures, exhibit excellent fluorescence properties with high quantum yield and a long lifetime. From an application perspective, exploring RE-based complexes in luminescent optoelectronic devices asks for effective modulation approaches that control the properties. Here we report electrically modulated phenomenon complex, namely Eu16(μ4-F)6(μ3-F)12(tBuCOO)18[N(CH2CH2O)3]4 (EuFC-16) particles, which effectively controls...
Abstract The evolution and driving factors underlying dust activity in central Asia remain controversial, particularly its effects on downwind regions. We present a Holocene storm record retrieved from the Tarim Basin (TB) perform linear nonlinear analyses records TB Greenland areas. results indicate similar response of activities to total solar irradiance both areas, an outbreak storms at ~3.5 kyr BP. suggest that decreasing temperature high northern latitudes, aided by change, reached...
Abstract Fast late Miocene global cooling since ∼7 Ma accompanied by less changeable atmospheric CO 2 levels revealed existing proxy reconstructions has suggested an intriguing tectonic‐climate link that remains controversial. Here, we present Cenozoic clay mineral records of the silicate weathering intensity from Chinese Loess Plateau and northeastern Tibetan to demonstrate a remarkable increase in at ∼9–7 induced enhanced monsoon. This change caused consumption ranging 0.18 1.8 × 10 11 mol...
Modern deep neural networks for classification usually jointly learn a backbone representation and linear classifier to output the logit of each class. A recent study has shown phenomenon called collapse that within-class means features vectors converge vertices simplex equiangular tight frame (ETF) at terminal phase training on balanced dataset. Since ETF geometric structure maximally separates pair-wise angles all classes in classifier, it is natural raise question, why do we spend an...