Binary kernel logistic regression: A sparsity-inducing formulation and a convergent decomposition training algorithm

Hebbian learning limits Hopfield network storage capacity (pattern-to-neuron ratio around 0.14). We propose Kernel Logistic Regression (KLR) learning. Unlike linear methods, KLR uses kernels to implicitly map patterns to high-dimensional feature space, enhancing separability. By learning dual variables, KLR dramatically improves storage capacity, achieving perfect recall even when pattern numbers exceed neuron numbers (up to ratio 1.5 shown), and enhances noise robustness. KLR demonstrably outperforms Hebbian and linear logistic regression approaches.

4187 Background: The PREDICT trial, a recent phase IIIb/IV study, aimed to address the critical need for improved personalized treatment strategies in advanced pancreatic cancer. This translational investigation examined the predictive value of 1 st line chemotherapy (CTX) response on the efficacy of subsequent 2 nd line treatment by using liquid biomarkers combined with machine learning (ML). Methods: Pts. were stratified into two cohorts based on short or long 2 nd line CTX time to treatment failure (S-/L-TTF2; n = 10 per group, 80/20% quantiles). Treatment-naïve PDAC tissue specimens underwent laser microdissection for tumor cell enrichment, followed by RNA profiling using NanoString™ PanCancer IO360. Selected differentially regulated genes from the omics results were utilized to screen peripheral blood mononuclear cells (PBMCs) from 2 nd line treatment-naïve patients at protein (flow cytometry, FC) or RNA (RT-qPCR) levels. FC data was analyzed using R-based single-cell clustering (x-Shift, FlowSOM, T-REX) to generate HyperGates (HGs), with subsequent ML-based back-gating (HyperFinder algorithm) of the most differential clusters. Additionally, classical ManualGates (MGs) were generated. Feature selection employed the Weka-based WrapperSubsetEval (WSE) algorithm with eight different classifiers (NB, KLR, LR, SMO, IBk, RT, RF, J48). The classifier/subset with optimal performance was utilized for binary classification (S-TTF2 vs. L-TTF2) of training (80%, n = 66) and validation (20%, n = 16) datasets. Results: Transcriptome analysis of L- vs. S-TTF2 tumor tissues revealed increased inflammation (upregulated 18-gene signature), immune cell activation/infiltration (e.g. CD4/CD8 T cells), and immune exhaustion (upregulation of e.g. PDCD1, LAG3, CTLA4, TIGIT) in L-TTF tumors. Further, the favorable Bailey immunogenic subtype was enriched in the L-TTF2 group. Analysis of eight FC panels (19 candidates) revealed 1198 differential clusters with HGs and 881 classical MGs. Feature selection, combining FC data with RT-qPCR results and clinical parameters, identified a best performing signature of 5 HGs and 2 MGs for 7 protein markers (CXCR4, CD8, CD4, CD62P, CD307b, CD45, CD121b). ML using a kernel logistic regression successfully predicted S- and L-TTF2 binary groups prior to 2 nd line CTX with nal-IRI/5-FU/LV (ROC-AUC > 0.90 for training and validation). Conclusions: We identified a favorable tumor immune microenvironment in L-TTF aPDAC patients, characterized by CD8 T cell-inflamed ("hot") tumor tissues prior to 2 nd line CTX. A 7-marker liquid biomarker panel, comprising 7 flow cytometry PBMC population gates, was developed for early prediction of 2 nd line nal-IRI/5-FU/LV CTX success. These findings aim to advance personalized treatment strategies. Clinical trial information: NCT03468335 .

QEKLR: quantum-enhanced kernel logistic regression for classification

<p>Automatic detection of anti-patterns from source code can reduce software maintenance costs massively. Nowadays, machine learning approaches are very commonly used to identify anti-patterns. Hence, it is very crucial to choose a classifier that can be useful for detecting anti-patterns. This work aims to help practitioners to choose a suitable classifier to detect anti-patterns. In this paper, we highlight 16 classifiers in four different categories to detect anti-patterns. Furthermore, the performance of these classifiers is identified with the data pre-processing (DPP) to detect four commonly occurring anti-patterns from the three commonly used open-source Java projects’ source code. The accuracy of Dagging classifiers is 98.4%. Kernel logistic regression (KLR) also performs well i.e., 97%. In the case of time complexity, naive Bayes (NB), decision trees (DT), support vector machines (SVM), library for support vector machines (LibSVM), logistic, and LightGBM (LB) have less time complexity to build a model in all the projects.</p>

755 Background: The phase II NEONAX trial examined PO and A gemcitabine/nab-paclitaxel (G/nP) efficacy for rPDAC pts with a comprehensive biomarker program. This translational study aimed at identifying liquid biomarker for early prediction of G/nP success in each treatment group (PO and A) and for the combined group using machine learning (ML). Methods: Pts from both study arms were stratified into groups based on Short- or Long-DFS ( n=80 pts total; PO n=42 pts; A n=38 pts). Blood plasma from selected G/nP naïve pts was analyzed by multiplexed ELISA (mELISA, 80-Plex ProcartaPlex Human Immune Response) and mass spectrometry (MS; feature list generation) and clinical data were acquired for each patient. For ML pts were divided into training (80%) and validation (20%) datasets. The Weka-based algorithm WrapperSubsetEval (WSE) with 8 different classifiers was used for feature selection. The best performing (highest accuracy, best ROC-AUC, minimal signature) classifier with the corresponding feature panel was selected by a 10x10-fold cross-validation (CV). Respective panels for all features combined as compared to clinical features and Ca19-9 blood levels were tested for performance via CV and bootstrap aggregating for training and validation datasets (10x with replacement). Results: The feature generation process generated 579 features from G/nP-naive pts (mELISA: 80; clinical data: 99; MS: 400). For response prediction of the whole rPDAC group to G/nP (S-/L-DFS), the ML-based flow identified a panel of 8 proteins from MS (SERPINA1, C1QB, KRT1, C4B, UBB, VCAM1, HPD, LYZ) and 3 proteins from mELISA (Galectin3, IL34, CCL4) combined with a logistic regression classifier. The predictive panel with the best performance for determining PO G/nP success was established using a kernel logistic regression classifier and included 2 proteins from mELISA (CXCL2, IL17A), 4 proteins from MS (C3, IGLV3-25, HLA-B, SHBG), and 2 clinical data (WHO grade, Na/Mg blood level ratio). The best performing biomarker for predicting success in A included 2 clinical data (tumor size at staging, hematocrit) and 1 protein from mELISA (IL22) and was established with a random forest classifier. All panels represented minimal feature signatures (rPDAC: 11; PO: 8; A; 3) and showed a high performance with ROC-AUC (training) > 0.90, ROC-AUC (CV) > 0.85, and ROC-AUC (validation) > 0.90. All panels described were superior compared to similar predictive signatures (with corresponding ML classifiers) for all clinical data or Ca19-9 blood levels. Conclusions: We show that minimal liquid biomarker signatures for early prediction of G/nP success in the NEONAX trial based on mELISA, MS and clinical data can be established by ML. The study shows the potential of ML for biomarker panel development and the value of mELISA/MS for generation of feature lists to use in ML. Clinical trial information: NCT02047513 .

Minimally invasive diagnostics based on liquid biopsy makes it possible early detection of lung cancer (LC). The blood plasma circulating cell-free DNA (cfDNA) fragments reflect the genome and chromatin status and are considered as integral cancer biomarkers and the biological entities for 'cancer-of-origin' prediction. The aim of this work is to create a method for processing next-generation sequencing (NGS) data and an interpretable binary classification model (CM), which analyzed cfDNA fragmentation features for distinguishing healthy subjects and subjects with LC. 148 healthy subjects and 138 subjects with LC were included in the study. cfDNA fractions, isolated from blood plasma biospecimens, were used for DNA libraries preparations and NGS on the NovaSeq 6,000 Illumina system with a coverage of 100 million reads/sample. Twelve variables, describing the abundance and length distribution of cfDNA fragments within each genomic interval, and 40 variables based on the values of position-weight matrices, describing combinations of 5-bp-long terminal motifs of cfDNA fragments, were used to characterize genomic fragmentation. Classification models of the first phase of machine learning were based either on logistic regression with L1- and L2-regularization or were probabilistic CMs based on Gaussian processes. The second phase CM was based on kernel logistic regression. The final CM can distinguish healthy subjects and subjects with LC with AUC values of 0.872-0.875. The performance of developed CM was evaluated using datum and testing sets for each LC stage category. Sensitivity values ranged from 66.7 to 85.7%, from 77.8 to 100%, and from 70 to 80% for LC stages I, II, and III, respectively. Specificity values ranged from 79.3 to 90.0%. Thus, the CM has a good diagnostic value and does not require clinical or other data on tumor-associated biomarkers. The current method for LC detection has some advantages for future clinical implementation as a decision-making support system due to the performance of the CM requires data exclusively from NGS-analysis of blood plasma cfDNA fragmentation; the accuracy of the CM does not depend on any additional clinical data; the CM is highly interpretable and traceable; CM has appropriate modular architecture.

Binary Kernel Logistic Regression: A Sparsity-Inducing Formulation and a Convergent Decomposition Training Algorithm

Indonesia, the world’s 4th largest country with a population of 270 million in 2020, faces many challenges due to rapid population growth, including biodiversity loss and increased consumption of naturalresources. One of the cultural factors underlying the high rate of population growth is the perception of“Banyak Anak Banyak Rezeki“ that develops in the community. This study aims to identify and modelthe factors that influence the “Banyak Anak Banyak Rezeki” stigma and find solutions to overcome thisproblem. The research method used was quantitative, with a sample of 384 people in South Sulawesi,consisting of Bugis, Makassar, Toraja, and Mandar tribes. The variables studied include religiosity,tradition, number of children, and cognitive dissonance. The analysis techniques used were logisticregression (LR) and kernel logistic regression (KLR). The results showed that religiosity, number ofchildren, and cognitive dissonance had a significant effect on the “Banyak Anak Banyak Rezeki” stigma.The accuracy of the LR model reached 87.01% and increased to 93.51% after using KLR.

Scrutinizing gully erosion hotspots to predict gully erosion susceptibility using ensemble learning framework

Application of Naive Bayes, kernel logistic regression and alternation decision tree for landslide susceptibility mapping in Pengyang County, China

SCH: Symmetric Consistent Hashing for cross-modal retrieval

Accelerated discovery of oxidation-resistant ultra-high temperature ceramics via data driven methodology

Plant growth-promoting rhizobacteria (PGPRs) are bacteria that colonize the plant roots. These beneficial bacteria have an influence on plant development through multiple mechanisms, such as nutrient availability, alleviating biotic and abiotic stress, and secrete phytohormones. Therefore, their inoculation constitutes a powerful tool towards sustainable agriculture and crop production. To understand plant-PGPRs interaction we present the classification of PGPR using machine learning and meta-learning classifiers namely Support Vector Machine (SVM), Kernel Logistic Regression (KLR), meta-SVM and meta-KLR to predict the presence of Bacillus megaterium inoculated in tomato root tissues using publicly available transcriptomic data. The original dataset presents 36 significantly differentially expressed genes. As the meta-KLR achieved near-optimal performance considering all the relevant metrics, this meta learner was afterwards used to identify the informative genes (IGs). The outcomes showed 157 IGs, being present all significantly differentially expressed genes previously identified. Among the IGs, 113 were identified as tomato genes, 5 as Bacillus subtilis proteins, 1 as Escherichia coli protein and 6 were unidentified. Then, a functional enrichment analysis of the tomato IGs showed 175 biological processes, 22 molecular functions and 20 KEGG pathways involved in B. megaterium–tomato interaction. Furthermore, the biological networks study of their Arabidopsis thaliana orthologous genes identified the co-expression, predicted interaction, shared protein domains and co-localization networks.

Basing on the machine learning model to analyse the coronary calcification score and the coronary flow reserve score to evaluate the degree of coronary artery stenosis

Learning non-parametric kernel via matrix decomposition for logistic regression

Modelling, mapping and monitoring of forest cover changes, using support vector machine, kernel logistic regression and naive bayes tree models with optical remote sensing data

Quantum kernel logistic regression based Newton method

Correction to: Modeling landslide susceptibility using data mining techniques of kernel logistic regression, fuzzy unordered rule induction algorithm, SysFor and random forest

Kernel reconstruction learning