Search for: All records

Chest X-rays are commonly used for diagnosing and characterizing lung diseases, but the complex morphological patterns in radiographic appearances can challenge clinicians in making accurate diagnoses. To address this challenge, various learning methods have been developed for algorithm-aided disease detection and automated diagnosis. However, most existing methods fail to account for the heterogeneous variability in longitudinal imaging records and the presence of missing or inconsistent temporal data. In this paper, we propose a novel longitudinal learning framework that enriches inconsistent imaging data over sequential time points by leveraging 2D Principal Component Analysis (2D-PCA) and a robust adaptive loss function. We also derive an efficient solution algorithm that ensures both objective and sequence convergence for the non-convex optimization problem. Our experiments on the CheXpert dataset demonstrate improved performance in capturing indicative abnormalities in medical images and achieving satisfactory diagnoses. We believe that our method will be of significant interest to the research community working on medical image analysis. 

The COVID-19 pandemic caused by SARS-CoV-2 has emphasized the importance of studying virus-host protein-protein interactions (PPIs) and drug-target interactions (DTIs) to discover effective antiviral drugs. While several computational algorithms have been developed for this purpose, most of them overlook the interplay pathways during infection along PPIs and DTIs. In this paper, we present a novel multipartite graph learning approach to uncover hidden binding affinities in PPIs and DTIs. Our method leverages a comprehensive biomolecular mechanism network that integrates protein-protein, genetic, and virus-host interactions, enabling us to learn a new graph that accurately captures the underlying connected components. Notably, our method identifies clustering structures directly from the new graph, eliminating the need for post-processing steps. To mitigate the detrimental effects of noisy or outlier data in sparse networks, we propose a robust objective function that incorporates the L2,p-norm and a constraint based on the pth-order Ky-Fan norm applied to the graph Laplacian matrix. Additionally, we present an efficient optimization method tailored to our framework. Experimental results demonstrate the superiority of our approach over existing state-of-the-art techniques, as it successfully identifies potential repurposable drugs for SARS-CoV-2, offering promising therapeutic options for COVID-19 treatment. 

Graphical representations are essential for comprehending high-dimensional data across diverse fields, yet their construction often presents challenges due to the limitations of traditional methods. This paper introduces a novel methodology, Beyond Simplex Sparse Representation (BSSR), which addresses critical issues such as parameter dependencies, scale inconsistencies, and biased data interpretation in constructing similarity graphs. BSSR leverages the robustness of sparse representation to noise and outliers, while incorporating deep learning techniques to enhance scalability and accuracy. Furthermore, we tackle the optimization of the standard simplex, a pervasive problem, by introducing a transformative approach that converts the constraint into a smooth manifold using the Hadamard parametrization. Our proposed Tangent Perturbed Riemannian Gradient Descent (T-PRGD) algorithm provides an efficient and scalable solution for optimization problems with standard simplex or L1-norm sphere constraints. These contributions, including the BSSR methodology, robustness and scalability through deep representation, shift-invariant sparse representation, and optimization on the unit sphere, represent major advancements in the field. Our work offers novel perspectives on data representation challenges and sets the stage for more accurate analysis in the era of big data. 

Linear discriminant analysis (LDA) is widely used for dimensionality reduction under supervised learning settings. Traditional LDA objective aims to minimize the ratio of squared Euclidean distances that may not perform optimally on noisy data sets. Multiple robust LDA objectives have been proposed to address this problem, but their implementations have two major limitations. One is that their mean calculations use the squared l2-norm distance to center the data, which is not valid when the objective does not use the Euclidean distance. The second problem is that there is no generalized optimization algorithm to solve different robust LDA objectives. In addition, most existing algorithms can only guarantee the solution to be locally optimal, rather than globally optimal. In this paper, we review multiple robust loss functions and propose a new and generalized robust objective for LDA. Besides, to better remove the mean value within data, our objective uses an optimal way to center the data through learning. As one important algorithmic contribution, we derive an efficient iterative algorithm to optimize the resulting non-smooth and non-convex objective function. We theoretically prove that our solution algorithm guarantees that both the objective and the solution sequences converge to globally optimal solutions at a sub-linear convergence rate. The experimental results demonstrate the effectiveness of our new method, achieving significant improvements compared to the other competing methods. 

Abstract. During the mid-Pliocene warm period (mPWP; 3.264–3.025 Ma), atmospheric CO2 concentrations were approximately 400 ppm, and the Antarctic Ice Sheet was substantially reduced compared to today. Antarctica is surrounded by the Southern Ocean, which plays a crucial role in the global oceanic circulation and climate regulation. Using results from the Pliocene Model Intercomparison Project (PlioMIP2), we investigate Southern Ocean conditions during the mPWP with respect to the pre-industrial period. We find that the mean sea surface temperature (SST) warming in the Southern Ocean is 2.8 °C, while global mean SST warming is 2.4 °C. The enhanced warming is strongly tied to a dramatic decrease in sea ice cover over the mPWP Southern Ocean. We also see a freshening of the ocean (sub)surface, driven by an increase in precipitation over the Southern Ocean and Antarctica. The warmer and fresher surface leads to a highly stratified Southern Ocean that can be related to weakening of the deep abyssal overturning circulation. Sensitivity simulations show that the decrease in sea ice cover and enhanced warming is largely a consequence of the reduction in the Antarctic Ice Sheet. In addition, the mPWP geographic boundary conditions are responsible for approximately half of the increase in mPWP SST warming, sea ice loss, precipitation, and stratification increase over the Southern Ocean. From these results, we conclude that a strongly reduced Antarctic Ice Sheet during the mPWP has a substantial influence on the state of the Southern Ocean and exacerbates the changes that are induced by a higher CO2 concentration alone. This is relevant for the long-term future of the Southern Ocean, as we expect melting of the western Antarctic Ice Sheet in the future, an effect that is not currently taken into account in future projections by Coupled Model Intercomparison Project (CMIP) ensembles. 

Accurate understanding of permafrost dynamics is critical for evaluating and mitigating impacts that may arise as permafrost degrades in the future; however, existing projections have large uncertainties. Studies of how permafrost responded historically during Earth’s past warm periods are helpful in exploring potential future permafrost behavior and to evaluate the uncertainty of future permafrost change projections. Here, we combine a surface frost index model with outputs from the second phase of the Pliocene Model Intercomparison Project to simulate the near‐surface (~3 to 4 m depth) permafrost state in the Northern Hemisphere during the mid-Pliocene warm period (mPWP, ~3.264 to 3.025 Ma). This period shares similarities with the projected future climate. Constrained by proxy-based surface air temperature records, our simulations demonstrate that near‐surface permafrost was highly spatially restricted during the mPWP and was 93 ± 3% smaller than the preindustrial extent. Near‐surface permafrost was present only in the eastern Siberian uplands, Canadian high Arctic Archipelago, and northernmost Greenland. The simulations are similar to near‐surface permafrost changes projected for the end of this century under the SSP5-8.5 scenario and provide a perspective on the potential permafrost behavior that may be expected in a warmer world. 

Abstract. Understanding the dominant climate forcings in the Pliocene is crucial to assessing the usefulness of the Pliocene as an analogue for our warmer future. Here, we implement a novel yet simple linear factorisation method to assess the relative influence of CO2 forcing in seven models of the Pliocene Model Intercomparison Project Phase 2 (PlioMIP2) ensemble. Outputs are termed “FCO2” and show the fraction of Pliocene climate change driven by CO2. The accuracy of the FCO2 method is first assessed through comparison to an energy balance analysis previously used to assess drivers of surface air temperature in the PlioMIP1 ensemble. After this assessment, the FCO2 method is applied to achieve an understanding of the drivers of Pliocene sea surface temperature and precipitation for the first time. CO2 is found to be the most important forcing in the ensemble forPliocene surface air temperature (global mean FCO2=0.56), sea surface temperature (global mean FCO2=0.56), and precipitation (global mean FCO2=0.51). The range between individual models is found to be consistent between these three climate variables, and the models generally show good agreement on the sign of the most important forcing. Our results provide the most spatially complete view of the drivers ofPliocene climate to date and have implications for both data–modelcomparison and the use of the Pliocene as an analogue for the future. ThatCO2 is found to be the most important forcing reinforces thePliocene as a good palaeoclimate analogue, but the significant effect ofnon-CO2 forcing at a regional scale (e.g. orography and ice sheet forcing at high latitudes) reminds us that it is not perfect, and these additional influencing factors must not be overlooked. This comparison is further complicated when considering the Pliocene as a state in quasi-equilibrium with CO2 forcing compared to the transient warming being experienced at present. 

Nonnegative Matrix Factorization (NMF) is broadly used to determine class membership in a variety of clustering applications. From movie recommendations and image clustering to visual feature extractions, NMF has applications to solve a large number of knowledge discovery and data mining problems. Traditional optimization methods, such as the Multiplicative Updating Algorithm (MUA), solves the NMF problem by utilizing an auxiliary function to ensure that the objective monotonically decreases. Although the objective in MUA converges, there exists no proof to show that the learned matrix factors converge as well. Without this rigorous analysis, the clustering performance and stability of the NMF algorithms cannot be guaranteed. To address this knowledge gap, in this article, we study the factor-bounded NMF problem and provide a solution algorithm with proven convergence by rigorous mathematical analysis, which ensures that both the objective and matrix factors converge. In addition, we show the relationship between MUA and our solution followed by an analysis of the convergence of MUA. Experiments on both toy data and real-world datasets validate the correctness of our proposed method and its utility as an effective clustering algorithm. 

Abstract Despite tectonic conditions and atmospheric CO 2 levels ( pCO 2 ) similar to those of present-day, geological reconstructions from the mid-Pliocene (3.3-3.0 Ma) document high lake levels in the Sahel and mesic conditions in subtropical Eurasia, suggesting drastic reorganizations of subtropical terrestrial hydroclimate during this interval. Here, using a compilation of proxy data and multi-model paleoclimate simulations, we show that the mid-Pliocene hydroclimate state is not driven by direct CO 2 radiative forcing but by a loss of northern high-latitude ice sheets and continental greening. These ice sheet and vegetation changes are long-term Earth system feedbacks to elevated pCO 2 . Further, the moist conditions in the Sahel and subtropical Eurasia during the mid-Pliocene are a product of enhanced tropospheric humidity and a stationary wave response to the surface warming pattern, which varies strongly with land cover changes. These findings highlight the potential for amplified terrestrial hydroclimate responses over long timescales to a sustained CO 2 forcing.