Accurate identification of proteins secreted into the bloodstream is essential for discovering diagnostic biomarkers and therapeutic targets. A significant challenge is the scarcity of experimentally validated blood-secretory proteins, limiting labeled datasets required for robust model training. To address this issue, we propose BloodProST, a novel machine-learning framework leveraging a self-training strategy to reliably predict blood-secretory proteins. BloodProST iteratively expands the labeled dataset by generating high-confidence pseudo-labels from a large pool of unlabeled protein sequences, thereby progressively enhancing model predictions without continuous manual annotation. At its core, BloodProST incorporates an unsupervised feature selection module based on differential evolution, optimizing the Silhouette score to identify the most discriminative physicochemical and sequence-derived features. Additionally, BloodProST employs a dual-pathway convolutional neural network and long short-term memory (CNN)-(LSTM) architecture: a CNN-based pathway captures local information from pre-constructed features, whereas an LSTM-based pathway extracts high-level sequential dependencies directly from protein sequences. Furthermore, domain-specific biological priors, such as the expected proportion of secretory proteins, are integrated into the model's loss function to guide training toward biologically plausible predictions. Extensive evaluation demonstrates that BloodProST significantly outperforms 14 state-of-the-art models across multiple metrics, achieving superior predictive accuracy, robustness, and interpretability. Validation analyses confirm the biological relevance of predictions through secretion-related markers (e.g. signal peptides and transmembrane regions) and demonstrate effective generalization to other biofluids, such as urine. Collectively, these results illustrate BloodProST's potential as a versatile computational tool for secretion prediction and biomarker discovery across diverse biological fluids.

Because of high dimensionality, correlation among covariates, and noise contained in data, dimension reduction (DR) techniques are often employed to the application of machine learning algorithms. Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and their kernel variants (KPC A, KLDA) are among the most popular DR methods. Recently, Supervised Kernel Principal Component Analysis (SKPCA) has been shown as another successful alternative. In this paper, brief reviews of these popular techniques are presented first. We then conduct a comparative performance study based on three simulated datasets, after which the performance of the techniques are evaluated through application to a pattern recognition problem in face image analysis. The gender classification problem is considered on MORPH-II and FG-NET, two popular longitudinal face aging databases. Several feature extraction methods are used, including biologically-inspired features (BIF), local binary patterns (LBP), histogram of oriented gradients (HOG), and the Active Appearance Model (AAM). After applications of DR methods, a linear support vector machine (SVM) is deployed with gender classification accuracy rates exceeding 95% on MORPH-II, competitive with benchmark results. A parallel computational approach is also proposed, attaining faster processing speeds and similar recognition rates on MORPH-II. Our computational approach can be applied to practical gender classification systems and generalized to other face analysis tasks, such as race classification and age prediction.

We perform preliminary studies on a large longitudinal face database MORPH-II, which is a benchmark dataset in the field of computer vision and pattern recognition. First, we summarize the inconsistencies in the dataset and introduce the steps and strategy taken for cleaning. The potential implications of these inconsistencies on prior research are introduced. Next, we propose a new automatic subsetting scheme for evaluation protocol. It is intended to overcome the unbalanced racial and gender distributions of MORPH-II, while ensuring independence between training and testing sets. Finally, we contribute a novel global framework for age estimation that utilizes posterior probabilities from the race classification step to compute a race-composite age estimate. Preliminary experimental results on MORPH-II are presented.

Automatic face age estimation is challenging due to its complexity owing to genetic difference, behavior and environmental factors, the dynamics of facial aging between different individuals, etc. In this work we propose to fuse the global facial feature extracted from Active Appearance Model (AAM) and the local facial features extracted from Local Binary Pattern (LBP), as the representation of faces. Furthermore, we introduce an advanced age estimation system combining feature fusion and model selection schemes such as Least Angle Regression (LAR) and sequential approaches. Due to the fact that different facial feature representations may come with various types of measurement scales, we compare multiple normalization schemes for both facial features. We demonstrate that the feature fusion with model selection can achieve significant improvement in age estimation over single feature representation alone. Our experiment on multi-ethnicity UIUC-PAL database suggests that age estimation with feature fusion and model selection outperforms the single feature, or the full feature model.