1. |
- Ramonet, M., et al.
(författare)
-
The fingerprint of the summer 2018 drought in Europe on ground-based atmospheric CO 2 measurements : Atmospheric CO 2 anomaly
- 2020
-
Ingår i: Philosophical Transactions of the Royal Society B: Biological Sciences. - : The Royal Society. - 0962-8436 .- 1471-2970. ; 375:1810
-
Tidskriftsartikel (refereegranskat)abstract
- During the summer of 2018, a widespread drought developed over Northern and Central Europe. The increase in temperature and the reduction of soil moisture have influenced carbon dioxide (CO 2) exchange between the atmosphere and terrestrial ecosystems in various ways, such as a reduction of photosynthesis, changes in ecosystem respiration, or allowing more frequent fires. In this study, we characterize the resulting perturbation of the atmospheric CO 2 seasonal cycles. 2018 has a good coverage of European regions affected by drought, allowing the investigation of how ecosystem flux anomalies impacted spatial CO 2 gradients between stations. This density of stations is unprecedented compared to previous drought events in 2003 and 2015, particularly thanks to the deployment of the Integrated Carbon Observation System (ICOS) network of atmospheric greenhouse gas monitoring stations in recent years. Seasonal CO 2 cycles from 48 European stations were available for 2017 and 2018. Earlier data were retrieved for comparison from international databases or national networks. Here, we show that the usual summer minimum in CO 2 due to the surface carbon uptake was reduced by 1.4 ppm in 2018 for the 10 stations located in the area most affected by the temperature anomaly, mostly in Northern Europe. Notwithstanding, the CO 2 transition phases before and after July were slower in 2018 compared to 2017, suggesting an extension of the growing season, with either continued CO 2 uptake by photosynthesis and/or a reduction in respiration driven by the depletion of substrate for respiration inherited from the previous months due to the drought. For stations with sufficiently long time series, the CO 2 anomaly observed in 2018 was compared to previous European droughts in 2003 and 2015. Considering the areas most affected by the temperature anomalies, we found a higher CO 2 anomaly in 2003 (+3 ppm averaged over 4 sites), and a smaller anomaly in 2015 (+1 ppm averaged over 11 sites) compared to 2018. This article is part of the theme issue 'Impacts of the 2018 severe drought and heatwave in Europe: from site to continental scale'.
|
|
2. |
- Loni, Mohammad, PhD Candidate, 1991-, et al.
(författare)
-
Learning Activation Functions for Sparse Neural Networks
- 2023
-
Ingår i: Proc. Mach. Learn. Res.. - : ML Research Press.
-
Konferensbidrag (refereegranskat)abstract
- Sparse Neural Networks (SNNs) can potentially demonstrate similar performance to their dense counterparts while saving significant energy and memory at inference. However, the accuracy drop incurred by SNNs, especially at high pruning ratios, can be an issue in critical deployment conditions. While recent works mitigate this issue through sophisticated pruning techniques, we shift our focus to an overlooked factor: hyperparameters and activation functions. Our analyses have shown that the accuracy drop can additionally be attributed to (i) Using ReLU as the default choice for activation functions unanimously, and (ii) Fine-tuning SNNs with the same hyperparameters as dense counterparts. Thus, we focus on learning a novel way to tune activation functions for sparse networks and combining these with a separate hyperparameter optimization (HPO) regime for sparse networks. By conducting experiments on popular DNN models (LeNet-5, VGG-16, ResNet-18, and EfficientNet-B0) trained on MNIST, CIFAR-10, and ImageNet-16 datasets, we show that the novel combination of these two approaches, dubbed Sparse Activation Function Search, short: SAFS, results in up to 15.53%, 8.88%, and 6.33% absolute improvement in the accuracy for LeNet-5, VGG-16, and ResNet-18 over the default training protocols, especially at high pruning ratios.
|
|
3. |
- de Haan, Anke, et al.
(författare)
-
Efficacy and moderators of efficacy of cognitive behavioural therapies with a trauma focus in children and adolescents: an individual participant data meta-analysis of randomised trials
- 2024
-
Ingår i: The Lancet Child and Adolescent Health. - 2352-4642. ; 8:1, s. 28-39
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Existing clinical trials of cognitive behavioural therapies with a trauma focus (CBTs-TF) are underpowered to examine key variables that might moderate treatment effects. We aimed to determine the efficacy of CBTs-TF for young people, relative to passive and active control conditions, and elucidate putative individual-level and treatment-level moderators. Methods: This was an individual participant data meta-analysis of published and unpublished randomised studies in young people aged 6−18 years exposed to trauma. We included studies identified by the latest UK National Institute of Health and Care Excellence guidelines (completed on Jan 29, 2018) and updated their search. The search strategy included database searches restricted to publications between Jan 1, 2018, and Nov 12, 2019; grey literature search of trial registries ClinicalTrials.gov and ISRCTN; preprint archives PsyArXiv and bioRxiv; and use of social media and emails to key authors to identify any unpublished datasets. The primary outcome was post-traumatic stress symptoms after treatment (<1 month after the final session). Predominantly, one-stage random-effects models were fitted. This study is registered with PROSPERO, CRD42019151954. Findings: We identified 38 studies; 25 studies provided individual participant data, comprising 1686 young people (mean age 13·65 years [SD 3·01]), with 802 receiving CBTs-TF and 884 a control condition. The risk-of-bias assessment indicated five studies as low risk and 20 studies with some concerns. Participants who received CBTs-TF had lower mean post-traumatic stress symptoms after treatment than those who received the control conditions, after adjusting for post-traumatic stress symptoms before treatment (b=−13·17, 95% CI −17·84 to −8·50, p<0·001, τ2=103·72). Moderation analysis indicated that this effect of CBTs-TF on post-traumatic stress symptoms post-treatment increased by 0·15 units (b=−0·15, 95% CI −0·29 to −0·01, p=0·041, τ2=0·03) for each unit increase in pre-treatment post-traumatic stress symptoms. Interpretation: This is the first individual participant data meta-analysis of young people exposed to trauma. Our findings support CBTs-TF as the first-line treatment, irrespective of age, gender, trauma characteristics, or carer involvement in treatment, with particular benefits for those with higher initial distress. Funding: Swiss National Science Foundation.
|
|
4. |
- Yver-kwok, Camille, et al.
(författare)
-
Evaluation and optimization of ICOS atmosphere station data as part of the labeling process
- 2021
-
Ingår i: Atmospheric Measurement Techniques. - : Copernicus GmbH. - 1867-8548 .- 1867-1381. ; 14:1, s. 89-116
-
Tidskriftsartikel (refereegranskat)abstract
- The Integrated Carbon Observation System (ICOS) is a pan-European research infrastructure which provides harmonized and high-precision scientific data on the carbon cycle and the greenhouse gas budget. All stations have to undergo a rigorous assessment before being labeled, i.e., receiving approval to join the network. In this paper, we present the labeling process for the ICOS atmosphere network through the 23 stations that were labeled between November 2017 and November 2019. We describe the labeling steps, as well as the quality controls, used to verify that the ICOS data (CO2, CH4, CO and meteorological measurements) attain the expected quality level defined within ICOS. To ensure the quality of the greenhouse gas data, three to four calibration gases and two target gases are measured: one target two to three times a day, the other gases twice a month. The data are verified on a weekly basis, and tests on the station sampling lines are performed twice a year. From these high-quality data, we conclude that regular calibrations of the CO2, CH4 and CO analyzers used here (twice a month) are important in particular for carbon monoxide (CO) due to the analyzer's variability and that reducing the number of calibration injections (from four to three) in a calibration sequence is possible, saving gas and extending the calibration gas lifespan. We also show that currently, the on-site water vapor correction test does not deliver quantitative results possibly due to environmental factors. Thus the use of a drying system is strongly recommended. Finally, the mandatory regular intake line tests are shown to be useful in detecting artifacts and leaks, as shown here via three different examples at the stations.
|
|