Optimal Dimensionality Reduction Research Articles

Helix Formation from Hydrogen Bond Kinetics in Alanine Homopeptides

We present an analysis of α-helix folding in the coarse-grained coordinate of number of formed helical hydrogen bonds (NHBs) for four alanine peptides (ALA)n, with n = 5, 8, 15, and 21 residues. Starting with multi-microsecond all-atom molecular dynamics trajectories in aqueous solution, we represent the system dynamics in a space of between four (for ALA5) and twenty (for ALA21) hydrogen-bonding microstates. In all cases, transitions changing the hydrogen bond count by 1–2 dominate and the coil formation, NHB 1 → 0, is the fastest process. The calculation of global maximum weight paths shows that, when analyzed at a sufficiently long lag time, folding in the NHB coordinate is consecutive, with direct folding, 0 → 3, for ALA5 and bottlenecks at transitions 4 → 6 for ALA8, 0 → 5 for ALA15, and 0 → 9 for ALA21. Further coarse-graining to 2–4 dimensions was performed with the optimal dimensionality reduction method, allowing the identification of crucial folding intermediates and time scales of their formation in ALA8, ALA15, and ALA21. The detailed analysis of hydrogen bonding patterns revealed that folding is initiated preferentially at both peptide termini. The kinetic model was also used to estimate diffusion and friction coefficients for helix propagation. The description of the helix formation process in the hydrogen bonding coordinate NHB was in good general agreement with the experimental data and qualitatively similar to previous kinetic models of higher dimensions based on structural clustering. Use of the low-dimensional hydrogen bonding picture thus provides a different, complementary way of describing the complex and fascinating mechanism of helix formation.

Integrated machine learning-driven disulfidptosis profiling: CYFIP1 and EMILIN1 as therapeutic nodes in neuroblastoma

BackgroundNeuroblastoma (NB), a prevalent pediatric solid tumor, presents formidable challenges due to its high malignancy and intricate pathogenesis. The role of disulfidptosis, a novel form of programmed cell death, remains poorly understood in the context of NB.MethodsGaussian mixture model (GMM)-identified disulfidptosis-related molecular subtypes in NB, differential gene analysis, survival analysis, and gene set variation analysis were conducted subsequently. Weighted gene co-expression network analysis (WGCNA) selected modular genes most relevant to the disulfidptosis core pathways. Integration of machine learning approaches revealed the combination of the Least absolute shrinkage and selection operator (LASSO) and Random Survival Forest (RSF) provided optimal dimensionality reduction of the modular genes. The resulting model was validated, and a nomogram assessed disulfidptosis characteristics in NB. Core genes were filtered and subjected to tumor phenotype and disulfidptosis-related experiments.ResultsGMM clustering revealed three distinct subtypes with diverse prognoses, showing significant variations in glucose metabolism, cytoskeletal structure, and tumor-related pathways. WGCNA highlighted the red module of genes highly correlated with disulfide isomerase activity, cytoskeleton formation, and glucose metabolism. The LASSO and RSF combination yielded the most accurate and stable prognostic model, with a significantly worse prognosis for high-scoring patients. Cytological experiments targeting core genes (CYFIP1, EMILIN1) revealed decreased cell proliferation, migration, invasion abilities, and evident cytoskeletal deformation upon core gene knockdown.ConclusionsThis study showcases the utility of disulfidptosis-related gene scores for predicting prognosis and molecular subtypes of NB. The identified core genes, CYFIP1 and EMILIN1, hold promise as potential therapeutic targets and diagnostic markers for NB.

Advanced Plant Disease Segmentation in Precision Agriculture using Optimal Dimensionality Reduction with Fuzzy C-Means Clustering and Deep Learning

Advanced Plant Disease Segmentation in Precision Agriculture using Optimal Dimensionality Reduction with Fuzzy C-Means Clustering and Deep Learning

ShadowAQP: Efficient Approximate Group-by and Join Query via Attribute-Oriented Sample Size Allocation and Data Generation

Approximate query processing (AQP) is one of the key techniques to cope with big data querying problem on account that it obtains approximate answers efficiently. To address non-trivial sample selection and heavy sampling cost issues in AQP, we propose ShadowAQP, an efficient and accurate approach based on attribute-oriented sample size allocation and data generation. We select samples according to group-by and join attributes, and determine the sample size for each group of unique value combinations to improve query accuracy. We design a conditional variational autoencoder model with automatic table data encoding and model update strategies. To further improve accuracy and efficiency, we propose a set of extensions, including parallel multi-round sampling aggregation, data outlier-aware sampling, and dimension reduction optimization. Evaluation results on diversified datasets show that, compared with SOTA approaches, ShadowAQP achieves 5.8× query speed performance improvement on average (up to 12.8×), while reducing query error by 74% on average (up to 95%) at the same time.

GB1 hairpin kinetics: capturing the folding pathway with molecular dynamics, replica exchange and optimal dimensionality reduction

We have performed molecular dynamics (MD) and replica-exchange (REMD) simulations of folding of the GB1 hairpin peptide in aqueous solution. REMD results were consistent with a cooperative zipper folding model. 120 MD trajectories at 320 K yielded relaxation times of 1.8 and 100 ns, with the slower assigned to global folding. The MD folding/unfolding transitions also followed the cooperative zipper model, specifying nucleation at the central turn followed by consecutive hydrogen bond formation. Formation of hydrogen bonds and hydrophobic contacts were highly correlated. Coarse-grained kinetic models constructed with the Optimal Dimensionality Reduction (ODR) approach found a folding time of 3.3 and unfolding time of 4.0 Additionally, relaxation times in the 130–170 ns range could be assigned to formation of the transition state and off-path intermediates. The unfolded state was the most highly populated and, significantly, most heterogenous, assembling the largest number of microstates, primarily composed of extended and turn structures. The folded state was also heterogenous, but a to a lesser degree, involving the fully folded and partially folded in-register hairpins at early stages of the zipper pathway. The transition state corresponded to the nucleated hairpin, with central turn and first beta-sheet hydrogen bond, while the off-path intermediates were off-register partial hairpins. Our simulation results were in excellent agreement with experimental data on folded fraction, relaxation time and folding mechanism. The new findings from this work suggest a highly cooperative zipper folding mechanism, nascent hairpin transition state and ∼100 ns relaxation related to intermediate formation. Communicated by Ramaswamy H. Sarma

Medical Image Fusion Using CNN with Automated Pooling

Background/Objectives : The purpose of this study is to demonstrate that the most appropriate pooling is selected among different types of pooling techniques in CNN, for different applications related to image fusion. The work presents a novel method by which pooling is selected automatically as per the needs of the application. Methods: This study presents the use of multiple functions for pooling techniques which are selected automatically based on the input images to produce fused optimum output. This will make the classification process much easier and better end results are obtained. Finding: The developed automated pooling method performs better concerning processing time and produces optimal output for parameters like Peak Noise Signal Ratio, Mean Square Error, Fusion Factor, Fusion Fidelity, and Visual Information Fidelity in comparison to existing methods like Discrete Wavelet Transformation, Non-Subsampled Contourlet Transform, and Principal Component Analysis. Novelty: The novel technique presented in this paper for automated selection of pooling method provides optimal dimension reduction in both the phases of CNN, and hence allows CNN to converge faster and optimally. Keywords: Image Fusion; CNN; Machine Learning; Pooling; Medical Image Analysis

Unsupervised machine learning improves risk stratification of patients with visual normal SPECT myocardial perfusion imaging assessments

Abstract Background Unsupervised machine learning has the potential to identify new cardiovascular phenotypes and more accurately assess individual risk in an unbiased fashion. Purpose We aimed to use unsupervised learning to identify, analyze, and risk-stratify subgroups of patients with normal perfusion by visual interpretation on single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI). Methods We included consecutive patients with visual normal clinical assessment (summed stress score of 0) from the multicenter (9 sites), REFINE SPECT registry. We considered 23 clinical, 17 image-acquisition, and 26 imaging variables. Optimal dimensionality reduction (Uniform Manifold Approximation and Projection), clustering (Gaussian Mixture Model), and number of clusters were selected to maximize the silhouette coefficient (how similar a patient is to those in their own cluster compared to other clusters). Risk stratification for all-cause mortality (ACM) and major adverse cardiac events (MACE) was assessed within these clusters and compared to risk stratification by quantitative ischemia (<5%, 5–10%, >10%) using Kaplan-Meier curves and Cox Proportional-Hazards analysis. Results In total, 17,527 (of 30,351) patients in the registry had visually normal perfusion, 49.7% female, median age of 64 [55, 72] years. There were 1,138 ACM events and 2,091 MACE events with a median follow-up of 4.1 [2.9, 5.7] years. Unsupervised learning provided better risk stratification for both ACM and MACE compared to quantitative ischemia (Figure). Notably, the high-risk cluster by unsupervised learning had a hazard ratio (HR) of 9.5 (95% confidence interval [CI]: 7.7–11.7) compared to 1.4 (95% CI: 1.1–1.9) for quantitative ischemia >10%. The high-risk cluster had proportionally more women (45% [low-risk], 51% [medium-risk], 57% [high-risk], all p<0.001), higher body mass indices (26.9, 27.4, 29.6, all p<0.001), prevalence of diabetes (17%, 22%, 33%, all p<0.001), and abnormal rest ECGs (30%, 43%, 64%, p<0.001); with lower rates of family history of coronary artery disease (40%, 33%, 24%, p<0.001). Patients in the low-risk cluster were more likely to undergo exercise stress (100%, 38%, 0%, all p<0.001), had lower rest peak systolic blood pressure (130, 131, 140 mmHg, all p<0.001), and higher stress peak systolic blood pressure (164, 150, 131 mmHg, all p<0.001). Patients in the high-risk cluster had higher left ventricular mass (129, 135.45, 143.9 g, all p<0.001) and stress volume (57, 59, 66 ml, all p<0.001). Conclusion Unsupervised learning identified new phenotypic clusters for SPECT MPI patients with visual normal assessments which provided improved risk stratification for ACM and MACE compared to SPECT ischemia. Such individualized risk assessment may allow better targeted management of patients with visually normal perfusion. Funding Acknowledgement Type of funding sources: Public Institution(s). Main funding source(s): Research reported in this publication was supported by the National Heart, Lung, And Blood Institute of the National Institutes of Health under Award Number R01HL089765. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Rapid Detection of Carbapenem-Resistant Klebsiella pneumoniae Using Machine Learning and MALDI-TOF MS Platform.

BackgroundRapid detection of carbapenem-resistant Klebsiella pneumoniae (CRKP) is essential for specific antimicrobial therapy. Machine learning techniques combined with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) can be used as a rapid, reliable, sensitive, and low-cost species identification method.MethodsClinically collected K. pneumoniae were subjected to MALDI-TOF MS analysis. A random forest (RF) algorithm and non-linear support vector machine (SVM) were used to construct the RF, SVM, and dimension reduction (SVM-K) models, and their performance was assessed for accuracy, sensitivity, specificity, and area under the subject worker curve (AUC).ResultsThe RF, SVM and SVM-K models showed good classification performance with 0.88, 0.88, and 0.91 accuracy, 0.82, 0.85, and 0.89 sensitivity, 0.93, 0.92, and 0.94 specificity with an AUC of 0.9013, 0.9298, and 0.9356, respectively. For the SVM-K model, the optimal dimension reduction was 105 to 153, and the average accuracy was >0.9. The top 10 peak features of significance according to the RF algorithm with 6515 Da appeared in 56.8% of CRKP isolates and 5.3% of CSKP isolates, which indicated the best classification performance.ConclusionThe three RF, SVM, and SVM-K models showed excellent classification performance differentiating the CRKP from CSKP; the SVM-K model was the best. Data analysis with machine learning combined with MALDI-TOF MS can be employed as a rapid and inexpensive alternative to existing detection methods.

Projected fuzzy C-means with probabilistic neighbors

In recent years, graph optimization dimensionality reduction methods have become a research hotspot in machine learning. The main challenge of these methods is how to choose proper neighbors for graph construction. For high-dimensional data clustering tasks, most methods often conduct a dimensionality reduction method at first and then perform a clustering method in sequence. However, such a sequential strategy may not be optimal because the reduced data obtained in the first stage may not be suitable for clustering. In this article, a novel method called Projected Fuzzy c-means with Probabilistic Neighbors(PFCM), which unifies graph optimization and Fuzzy c-means, is proposed. Our model projects the data into an optimal subspace at first and then learns the sparse weights matrix by considering probabilistic neighbors and membership matrix together on the projected data. The above two steps run iteratively until the algorithm converges. Especially, L0-norm constraints are employed on the weights matrix to avoid the obstacles caused by outliers. An optimization procedure is designed to solve the proposed model effectively. We conducted numerous experiments on eight benchmark data sets. The experimental results show that the performance of the proposed method is better than some available dimensionality reduction algorithms for clustering tasks.

Detection of moisture content in salted sea cucumbers by hyperspectral and low field nuclear magnetic resonance based on deep learning network framework

The accurate control of moisture content (MC) during the processing of sea cucumber is beneficial to improve the taste of sea cucumber and maintain its nutritional value, which is directly related to the quality and shelf life of sea cucumber. The purpose of this study is to explore the feasibility using deep learning (DL) to realize rapid nondestructive detection of MC in salted sea cucumbers based on hyperspectral imaging (HSI) and low field nuclear magnetic resonance (LF-NMR) data. Firstly, three Cuckoo Search (CS) dimensionality reduction algorithms (Traditional-CS, Binary-CS and Chaotic-CS) were combined with DL framework respectively using HSI and LF-NMR data to establish prediction models, which proved the feasibility of DL framework in predicting the MC of sea cucumbers, and Chaotic-CS algorithm was selected as the optimal dimensionality reduction algorithm. Then, the MC visualization based on HSI and LF-NMR data was realized respectively to detect the migration and decrease of MC. Finally, using both HSI and LF-NMR data, the advantages of the models based on Fusion-net DL (FDL) framework were discussed, which showed better performance than the single-data models, with RC2 of 0.9929, RMSEC of 0.0016, RP2 of 0.9936 and RPD of 12.5041. In summary, the rapid nondestructive detection of MC in salted sea cucumbers could be realized by HSI and LF-NMR data based on DL framework, and the advantage of data fusion detection based on FDL framework was verified.

An Optimal Dimension Reduction Strategy and Experimental Evaluation for Parkinson’s Disease Classification

An Optimal Dimension Reduction Strategy and Experimental Evaluation for Parkinson’s Disease Classification

An optimal dimension reduction strategy and experimental evaluation for Parkinson's disease classification

The amount of data streamed and generated through various healthcare systems is exponentially increasing day by day. Applying traditional data mining algorithms on this massive sized data to construct automated decision support systems is a tedious and time consuming task. In recent years, there has been increasing interest in the development of telediagnosis and telemonitoring systems for Parkinson's disease (PD). Parkinson's disease is a progressive neurodegenerative disease which affect the movement characteristics. PD patients commonly face vocal impairments during the early stages of the disease. This work proposes a computationally efficient method for dimension reduction and classification of healthcare related data. The devised framework is capable to deal with the data having discrete as well as continuous natured features. The experimental evaluation is performed on Parkinson's disease classification database (Sakar et al., 2018). The statistical performance metrices used are - validation and test accuracy, precision, recall, F1-score, etc. There will be computational complexity advantages when this reduced dimension data is further processed for modelling and building prediction system. In order to prove the optimality of proposed framework, comparative analysis is performed with the significant existing approaches.

Approaching Optimal Nonlinear Dimensionality Reduction by a Spiking Neural Network

This work deals with the presentation of a spiking neural network as a means for efficiently solving the reduction of dimensionality of data in a nonlinear manner. The underneath neural model, which can be integrated as neuromorphic hardware, becomes suitable for intelligent processing in edge computing within Internet of Things systems. In this sense, to achieve a meaningful performance with a low complexity one-layer spiking neural network, the training phase uses the metaheuristic Artificial Bee Colony algorithm with an objective function from the principals in the machine learning science, namely, the modified Stochastic Neighbor Embedding algorithm. To demonstrate this fact, complex benchmark data were used and the results were compared with those generated by a reference network with continuous-sigmoid neurons. The goal of this work is to demonstrate via numerical experiments another method for training spiking neural networks, where the used optimizer comes from metaheuristics. Therefore, the key issue is defining the objective function, which can relate optimally the information at both sides of the spiking neural network. Certainly, machine learning techniques have advanced in defining efficient loss functions that can become suitable objective function candidates in the metaheuristic training phase. The practicality of these ideas is shown in this article. We use MSE values for evaluating the relative quality of the results and also co-ranking matrices.

Length Dependent Folding Kinetics of Alanine-Based Helical Peptides from Optimal Dimensionality Reduction.

We present a computer simulation study of helix folding in alanine homopeptides (ALA)n of length n = 5, 8, 15, and 21 residues. Based on multi-microsecond molecular dynamics simulations at room temperature, we found helix populations and relaxation times increasing from about 6% and ~2 ns for ALA5 to about 60% and ~500 ns for ALA21, and folding free energies decreasing linearly with the increasing number of residues. The helix folding was analyzed with the Optimal Dimensionality Reduction method, yielding coarse-grained kinetic models that provided a detailed representation of the folding process. The shorter peptides, ALA5 and ALA8, tended to convert directly from coil to helix, while ALA15 and ALA21 traveled through several intermediates. Coarse-grained aggregate states representing the helix, coil, and intermediates were heterogeneous, encompassing multiple peptide conformations. The folding involved multiple pathways and interesting intermediate states were present on the folding paths, with partially formed helices, turns, and compact coils. Statistically, helix initiation was favored at both termini, and the helix was most stable in the central region. Importantly, we found the presence of underlying universal local dynamics in helical peptides with correlated transitions for neighboring hydrogen bonds. Overall, the structural and dynamical parameters extracted from the trajectories are in good agreement with experimental observables, providing microscopic insights into the complex helix folding kinetics.

An adjoint‐assisted multilevel multifidelity method for uncertainty quantification and its application to turbomachinery manufacturing variability

AbstractIn this work we propose, analyze, and demonstrate an adjoint‐based multilevel multifidelity Monte Carlo (MLMF) framework called FastUQ. The framework is based on the MLMF of Geraci et al. and uses the Inexpensive Monte Carlo (IMC) method of Ghate as low‐fidelity surrogate. The setup cost of IMC‐1 surrogate in FastUQ requires just the adjoint solution at the input mean whose computational cost is independent of the number of input uncertainties making it suitable for solving problems with a large number of uncertain parameters. We demonstrate the robustness of the method to quantify uncertainties in aerodynamic parameters due to surface variations caused by the manufacturing processes for a highly loaded turbine cascade. A stochastic model for surface variations on the cascade is proposed and optimal dimensionality reduction of model parameters is realised using goal‐based principal component analysis using adjoint sensitivities of multiple quantities of interest. The proposed method achieves a 70% reduction in computational cost in predicting the mean quantities such as total‐pressure loss and mass flow rate compared to the state‐of‐art MLMC method.

Optimal Statistical Inference in the Presence of Systematic Uncertainties Using Neural Network Optimization Based on Binned Poisson Likelihoods with Nuisance Parameters

Data analysis in science, e.g., high-energy particle physics, is often subject to an intractable likelihood if the observables and observations span a high-dimensional input space. Typically the problem is solved by reducing the dimensionality using feature engineering and histograms, whereby the latter allows to build the likelihood using Poisson statistics. However, in the presence of systematic uncertainties represented by nuisance parameters in the likelihood, an optimal dimensionality reduction with a minimal loss of information about the parameters of interest is not known. This work presents a novel strategy to construct the dimensionality reduction with neural networks for feature engineering and a differential formulation of histograms so that the full workflow can be optimized with the result of the statistical inference, e.g., the variance of a parameter of interest, as objective. We discuss how this approach results in an estimate of the parameters of interest that is close to optimal and the applicability of the technique is demonstrated with a simple example based on pseudo-experiments and a more complex example from high-energy particle physics.

Optimal dimensionality reduction of Markov chains using graph transformation.

Markov chains can accurately model the state-to-state dynamics of a wide range of complex systems, but the underlying transition matrix is ill-conditioned when the dynamics feature a separation of timescales. Graph transformation (GT) provides a numerically stable method to compute exact mean first passage times (MFPTs) between states, which are the usual dynamical observables in continuous-time Markov chains (CTMCs). Here, we generalize the GT algorithm to discrete-time Markov chains (DTMCs), which are commonly estimated from simulation data, for example, in the Markov state model approach. We then consider the dimensionality reduction of CTMCs and DTMCs, which aids model interpretation and facilitates more expensive computations, including sampling of pathways. We perform a detailed numerical analysis of existing methods to compute the optimal reduced CTMC, given a partitioning of the network into metastable communities (macrostates) of nodes (microstates). We show that approaches based on linear algebra encounter numerical problems that arise from the requisite metastability. We propose an alternative approach using GT to compute the matrix of intermicrostate MFPTs in the original Markov chain, from which a matrix of weighted intermacrostate MFPTs can be obtained. We also propose an approximation to the weighted-MFPT matrix in the strongly metastable limit. Inversion of the weighted-MFPT matrix, which is better conditioned than the matrices that must be inverted in alternative dimensionality reduction schemes, then yields the optimal reduced Markov chain. The superior numerical stability of the GT approach therefore enables us to realize optimal Markovian coarse-graining of systems with rare event dynamics.

Speech feature selection and emotion recognition based on weighted binary cuckoo search

In this paper, a hybrid system is proposed for speech emotion recognition (SER). The algorithm in this paper adopts a two-stage design concept. In the first stage, we use the ensemble learning model random forest algorithm to obtain the importance of each feature. We use Emo-DB for experimental comparison and find that the combination of the logistic regression algorithm and the WBCS algorithm achieves best results. Cross-training method is used to ensure the features adapt to various situations. Under 100 training sets, the sentiment classification results are satisfactory. The proposed method is more accurate than state-of-the-art intelligent optimization dimensionality reduction algorithms.

Fault detection and diagnosis of nonlinear dynamical processes through correlation dimension and fractal analysis based dynamic kernel PCA

A novel Dynamic Kernel PCA (DKPCA) method is developed for process monitoring in nonlinear dynamical systems. Classical DKPCA approaches still exhibit vague linearity assumptions to determine the number of principal components and to construct the dynamical structure. The optimal Static PCA (SPCA) and Dynamic PCA (DPCA) structures are constructed herein through the powerful theory of the nonlinear Fractal Dimension (FDim). While DKPCA offers a generic data-driven modelling of nonlinear dynamical systems, the fractal correlation dimension provides an intrinsic measure of the data complexity counting for the nonlinear dynamics and the chaotic behaviour. The proposed Fractal-based DKPCA (FDKPCA) integrates the two strategies to overcome SPCA/DPCA/DKPCA shortcomings, FDim allows verifying the degree of fitting and ensures optimal dimensionality reduction. The novel fault detection and diagnosis method is validated through seven applications using the Process Network Optimization (PRONTO) benchmark with real heterogeneous data, FDKPCA showed superior performance compared to contemporary approaches.

A surrogate-assisted multi-objective evolutionary algorithm with dimension-reduction for production optimization

Multi-objective optimization (MOO), which involves more than one conflicting objective to be optimized simultaneously, is expected to provide efficient and comprehensive reservoir management (RM) solutions. The multi-objective production optimization problems are considered to be expensive due to the difficulties and cost of operations. Surrogate-assisted evolutionary algorithms (SAEAs), which has proved to be an effective way to solve expensive problems, design computationally cheap function to approximate each objective function. Meanwhile, the optimization process involves a large number of decision variables. However, building a high-quality surrogate model has become difficult due to the “curse of dimensionality”. Base on characterization, an efficient multi-objective optimization framework called SA-RVEA-PCA is proposed to effectively deal with large-scale and computationally expensive simulation-based optimization problems, including three parts:1) Given a set of simulation results, a Gaussian process (GP) model based on Principal Component Analysis (PCA) for each objective function is trained so that the surrogate models can guide the optimization more accurately. 2) A reference vector guided evolutionary algorithm (RVEA) recently developed is employed as a multi-objective optimizer. 3) The information of uncertainty given by GP and angle-penalized penalized (APD) proposed in RVEA are used to update the surrogate models. To the best of our knowledge, the proposed algorithm is applied to a benchmark function, and two typical applications of MOO with synthetic reservoir models. Results show that the proposed method can provide more comprehensive and efficient RM with a higher convergence speed.