Explore the vast range of titles available.

Accurately projecting climate change and its impact is crucial for quantifying the risk of extreme events and developing effective adaptation strategies. However, future projections exhibit substantial uncertainties among Earth system models (ESMs). Notably, the latest phase of the Coupled Model Intercomparison Project includes some ‘hot’ ESMs with high climate sensitivity that exceed the likely range inferred from multiple lines of evidence, leading to a broader uncertainty range compared to previous CMIP phases. Although various uncertainty quantification and constraint methods have been proposed, they are not yet widely adopted. The approach of using an equal-weighted ensemble average for projections remains prevalent. Here we examine commonly used uncertainty quantification methods and constraint projection methods, describing their characteristics. Subsequently, taking extreme precipitation as a case, we constrain the range of projection uncertainty employing two weighing constraint methods and two emergent constraint methods. The results demonstrate that all methods effectively reduce the uncertainty in extreme precipitation projections. Specifically, the comprehensive constraints reduce the projection uncertainty by 26%–31% at the long-term future (2081–2100) under different scenarios. Therefore, we strongly recommend that attention should be paid to quantifying and constraining uncertainty when undertaking future projections of climate change and its impacts.

Clustering by fast searching and finding density peaks (DPC) method can rapidly identify the centers of clusters which have relatively high densities and high distances according to a decision graph. Various methods have been introduced to extend the DPC model over the past five years. DPC was originally presented as an unsupervised learning algorithm, and the thought of adding some prior information to DPC emerges as an alternative approach for improving its performance. It is extravagant to collect labeled data in real applications, and annotation of class labels is a nontrivial work, while pairwise constraint information is easier to get. Furthermore, the class label information can be converted into pairwise constraint information. Thus, we can take full advantage of pairwise constraints (or prior information) as much as possible. So this paper presents a new semi-supervised density peaks clustering algorithm (SSDPC) that uses constraint projection, which is flexible in loosening a few constraints over the learning stage. In the first stage, instances involving instance-level constraints and the remaining instances are concurrently projected to a lower dimensional data space led by the pairwise constraints, where viewing the distribution of data instances more clearly is available. Subsequently, traditional DPC is executed on the new lower dimensional dataset. Lastly, a few datasets from the Microsoft Research Asia Multimedia (MSRA-MM) image and UCI machine learning repository datasets are adopted in the experimental validation. The experimental results demonstrate that the proposed SSDPC achieves better performance than other three semi-supervised clustering algorithms.

Ensemble learning is an effective technique for the class-imbalance problem, and the key for obtaining a successful ensemble is to create individual base classifiers with high accuracy and diversity. In this paper, we propose a novel ensemble learning method via constraint projection and undersampling technique, constructing each base classifier through the following two steps: 1) constructing a set of pairwise constraints by undersampling examples from the minority/majority class set and learning a projection matrix from the pairwise constraint set and 2) undersampling the original training set to obtaining a new training set on which a base classifier is constructed in the new feature space defined by the projection matrix. For the first step, the projection matrix is mainly used to enhance the separability between the diverse class examples and thus to improve the performance of the base classifier, and the undersampling technique is used to create diverse sets of pairwise constraints to train diverse projection matrices, thus introducing diversity to base classifiers. For the second step, the undersampling technique aims to improve the performance of base classifiers on the minority class and further increase the diversity between the individual base classifiers. The experimental results show that the proposed method shows significantly better performance on the measures of recall, g-mean, f-measure and AUC than other state-of-the-art methods for 29 datasets with various data distributions and imbalance ratios.

We consider the general two-stage convex stochastic programs with discrete distribution, in which the risk measure is only assumed to be convex and monotonic, not necessarily to be coherent or have special structures. We propose a scenario decomposition framework which incorporates subgradient computation and the incremental constraint projection steps. The decomposition of the algorithm is based on the scenario-wise separability of these computations. We analyze the convergence and local rate of convergence of the proposed method under mild conditions. This method is further applied to a class of distributionally robust two-stage stochastic programs. Numerical results of a practical multi-product assembly model are reported to demonstrate the effectiveness of the proposed method.

This paper presents a novel technique for wide-angle fisheye lens calibration which requires neither metric information nor particular reference pattern. First, the fisheye imaging model with the interior Orientation parameters (IOPs)—principal point ( u 0 ,v 0 ), focal length f , aspect ratio λ and radial distortion coefficients ( k 1 , k 2 ), is established. Then, upon the fisheye imaging model and the parameter dependency between f and ( k 1 , k 2 ), the radial distortion projection ellipse constraint (RDPEC) for space lines in fisheye image is mathematically formulated to build a non-linear calibration model for high-precision estimation of the IOPs. In this step, parameter initialization based on the geometry of fisheye image outline ellipse (FIOE) is discussed as well. Finally, initial IOPs are further optimized though least square technique by taking the projection ellipse arcs of space lines in fisheye image as observation. The proposed calibration technique was tested on two kinds of fisheye images: (a) simulated image with a set of ground-truth IOPs, (b) internet images with unknown IOPs. Experimental results show that the calibration parameters in this paper are in the best agreement with the fisheye imaging model, compared with the ground-truth parameters and the parameters estimated by two state-of-the-art literature. Compared to that by a state-of-the-art CNN and the well-known software DXO, the proposed technique can enable a high-quality correction of fisheye images in different regions. This makes it very useful in application scenarios containing space lines, such as urban panorama surveillance, auto-parking and, robot navigation.

By using Hamilton improved circle, transversely and longitudinally cut up paper join together. By scanning image and image extraction techniques, the images of shredded paper are obtained. The process of shredding splicing is mainly divided into two parts, the image preprocessing and edge matching. Then the images conver into grey value matrixes and binarization are implemented on the image with appropriate threshold. The images of shredded paper are classified into nineteen lines. With the shortest Euclidean distance between the edge of shredding and Hamilton improved circle, shredded paper is transversely spliced. The test shows that accuracy is as high as 100%.

This study proposes an adaptive approximation-based control approach for non-linear pure-feedback systems in the presence of full state constraints. Completely non-affine non-linear functions are considered and assumed to be unknown. The dynamic surface design based on integral barrier Lyapunov functionals is provided to achieve both the desired tracking performance and the constraints satisfaction, in consideration of the full-state-constrained non-affine non-linearities. In this design procedure, simple sufficient conditions for choosing control gains, which can be checked off-line, are established to guarantee the feasibility of the controller. The function approximation technique is employed to estimate unknown non-linearities induced from the controller design procedure where the adaptive laws using the projection operator are designed to ensure the boundedness of the function approximators in the feasibility conditions. It is shown that all the signals in the closed-loop system are uniformly ultimately bounded and the tracking error converges to an adjustable neighbourhood of the origin while all state variables always remain in the constrained state space.

Marine low clouds strongly cool the planet. How this cooling effect will respond to climate change is a leading source of uncertainty in climate sensitivity, the planetary warming resulting from CO2 doubling. Here, we observationally constrain this low cloud feedback at a near-global scale. Satellite observations are used to estimate the sensitivity of low clouds to interannual meteorological perturbations. Combined with model predictions of meteorological changes under greenhouse warming, this permits quantification of spatially resolved cloud feedbacks. We predict positive feedbacks from midlatitude low clouds and eastern ocean stratocumulus, nearly unchanged trade cumulus and a near-global marine low cloud feedback of 0.19 ± 0.12 W m−2 K−1 (90% confidence). These constraints imply a moderate climate sensitivity (~3 K). Despite improved midlatitude cloud feedback simulation by several current-generation climate models, their erroneously positive trade cumulus feedbacks produce unrealistically high climate sensitivities. Conversely, models simulating erroneously weak low cloud feedbacks produce unrealistically low climate sensitivities.Marine low clouds cool the planet, but their response to warming is uncertain and dominates the spread in model-based climate sensitivities. Observational constraints suggest smaller cloud feedbacks than previously reported and imply a more moderate climate sensitivity.

Reliable projections of wildfire and associated socioeconomic risks are crucial for the development of efficient and effective adaptation and mitigation strategies. The lack of or limited observational constraints for modeling outputs impairs the credibility of wildfire projections. Here, we present a machine learning framework to constrain the future fire carbon emissions simulated by 13 Earth system models from the Coupled Model Intercomparison Project phase 6 (CMIP6), using historical, observed joint states of fire-relevant variables. During the twenty-first century, the observation-constrained ensemble indicates a weaker increase in global fire carbon emissions but higher increase in global wildfire exposure in population, gross domestic production, and agricultural area, compared with the default ensemble. Such elevated socioeconomic risks are primarily caused by the compound regional enhancement of future wildfire activity and socioeconomic development in the western and central African countries, necessitating an emergent strategic preparedness to wildfires in these countries. A new study develops a machine learning framework to observationally constrain CMIP6-simulated fire carbon emissions, finding a weaker increase in 21st-century global fires but higher increase in their socioeconomic risks than previously thought.

Serving as ‘the water tower of Asia’, the Tibetan Plateau (TP) supplies water resources to more than 1.4 billion people. It is warming more rapidly than the global average over the past decades, affecting regional hydrological cycle and ecosystem services. However, the anthropogenic (ANT) influence remains unknown. Here we assessed the human contribution to the observed TP warming based on coupled climate simulations and an optimal fingerprinting detection and attribution analysis. We show that the observed rapid warming on the TP (1.23 °C over 1961–2005) is attributable to human influence, and particularly, to the greenhouse gases with a contribution of 1.37 °C by the best estimate, which was slightly offset by anthropogenic aerosols. As the multi-model ensemble tends to underestimate the ANT warming trend, the constraint from the attribution results suggests an even warmer future on the TP than previously expected, implying further increased geohazard risks in the Asian water tower.