Imaging problems, including image reconstruction, restoration, enhancement, segmentation,classification, object detection, and secure image transmission, have traditionallybeen addressed using analytical methods derived from mathematical models and domainspecificknowledge. Regularization, filtering, optimization, geometric modeling, andstatistical estimation remain important because their assumptions and behavior can oftenbe understood and controlled explicitly. More recently, Machine Learning (ML) and DeepLearning (DL) methods have achieved impressive results by learning complex image representationsdirectly from large datasets. Current research increasingly seeks to combinethese two perspectives. Instead of treating analytical and learning-based methods as competingalternatives, hybrid approaches incorporate physical constraints, signal-processingprinciples, or application-specific knowledge into trainable systems. This can improve dataefficiency, interpretability, robustness, and computational performance while retaining therepresentational power of modern neural networks. [...]
In distributed flexible job shops, preventive maintenance is required to manage machine degradation, but competition for limited maintenance resources can delay production. This study develops a bi-objective model to minimize makespan and total energy consumption by integrating factory assignment, operation sequencing, machine and processing speed selection, imperfect preventive maintenance, and maintenance resource allocation. To solve this problem, a multi-objective memetic algorithm with constrained maintenance resources (MRC-MOMA) is proposed. The algorithm combines four-layer encoding with a decoder that generates maintenance requests according to machine degradation and schedules them under maintenance resource constraints. Computational experiments show that MRC-MOMA achieves better overall performance than the comparison algorithms in terms of hypervolume (HV) and inverted generational distance (IGD), with sensitivity analysis supporting this advantage within the tested parameter ranges. For the representative instances, increasing maintenance resource capacity substantially reduces maintenance waiting time and shortens makespan, but does not necessarily reduce total energy consumption.
Keeping a quantitative review current in a fast-growing applied field is first a screening problem. We present two algorithms for modelling biofloc aquaculture. The first, Floccus, is an openly archived, PRISMA 2020-compliant pipeline that removes duplicates in three layers (deterministic, embedding-based and large language model (LLM)-judged) and screens each record against eleven criteria under an asymmetric decision rule with an auditable trail. Applied to 22,056 records from four databases, it yielded 16,965 unique and 4883 eligible studies in 25 thematic clusters. Agreement with blind human labels was low (κ=0.107), of the same order as between two human reviewers applying very different inclusion thresholds (κ=0.105; 65% versus 18% of records judged relevant); it saved more screening work than keyword and single-prompt baselines, and an aquatic-science database added under 1% of eligible studies. The second algorithm assembles the corpus’s six model families into one multi-scale hierarchy coupled through the aeration-driven dissipation field, with a proposed operator converting floc size into active biomass through the Thiele effectiveness factor. Manual screening confirmed formal floc modelling is rare in applied biofloc work (22 of 141 records). Corpus-level recall remains moderate (0.45), and the coupled model awaits validation against tank data.
Large language models (LLMs) generate fluent Chinese, but controlling lexical difficulty remains challenging. This study examined a Python-orchestrated external-control workflow using one GPT model, two sequential prompting conditions (no-word-list and lexicon-aware), three post-generation revision strategies (word-level substitution, sentence-level rewriting, and downstream regeneration), and four Test of Chinese as a Foreign Language (TOCFL) levels (A1–B2). Automatic indicators covered 96 initial drafts and 288 revision outputs. For 24 selected tasks, two experts directly rated 180 text instances; 12 unchanged outputs inherited initial-draft ratings, yielding 192 condition-specific expert-evaluation records. Task-matched comparisons used 20,000 TOCFL-stratified task-cluster bootstrap resamples and Holm-adjusted tests; automatic–expert associations were analyzed using Spearman correlations with task-cluster confidence intervals. Lexicon-aware prompting was associated with fewer initial out-of-level words and less revision, although the prompting batches were sequential. Word-level substitution and sentence-level rewriting generally showed greater source-relative embedding resemblance, whereas downstream regeneration received higher expert ratings for several language-quality dimensions but not consistently for level appropriateness. Residual out-of-level word density showed a weak, inconclusive pooled association with level appropriateness. Whole-text cosine similarity was negatively associated with semantic plausibility overall but positively associated within word-level substitution. Automatic indicators and source-blind expert judgments thus provide complementary, non-interchangeable evidence.
Unsupervised feature selection is difficult when relevance must be inferred without labels, and redundancy, outliers, and sample-localized geometry can distort marginal criteria. This study presents Geometry-Supported Residual Feature Selection (GSR-FS), a fully label-blind algorithm and open-source Python library that represents each feature by its contribution to pairwise sample geometry. Structural relevance is quantified by Bhattacharyya affinity to the aggregate geometry of the remaining features, calibrated against within-feature permutations, and downweighted by an entropy-based Sample-Geometric Coverage term. A non-negative residual reconstruction criterion then penalizes candidates whose geometry is already represented by previously selected features. Automatic selection uses a feature-wise max-null permutation distribution, whereas fixed-cardinality selection follows the same greedy score without using target information. The released library additionally provides balanced pair subsampling, numerically equivalent residual solvers, source-integrity audits, publication guardrails, and diagnostic structural-group outputs. In a source-audited 30-dataset evidence panel, 29 datasets supported complete five-method comparisons. Mean balanced accuracy was 0.6031 for GSR-FS, compared with 0.6019 for SPEC, 0.5970 for MCFS, 0.5892 for MaxVariance, and 0.5810 for Laplacian Score. The Friedman test was not significant (p = 0.488), and no Holm-corrected pairwise comparison was significant. Additional robustness analyses showed that automatic GSR-FS achieved mean balanced accuracy of 0.6216 compared with 0.6069 for fixed-cardinality GSR-FS on the 30-dataset panel, without a significant paired difference (p = 0.1215). Further validation showed high-precision independent-null calibration at α = 0.05, with 25 of 500 datasets producing a non-empty subset (empirical rate of 0.050; exact 95% CI of 0.0326–0.0729). Controlled dependent-feature experiments showed that GSR-FS combined higher latent-group coverage with lower redundancy than Laplacian Score and MCFS in the examined structural-recovery scenarios, while a recent FGMRW-UFS comparator showed limited recovery under this specific criterion. An engineering stress test was completed successfully through 20,000 candidate features under explicitly capped pair and permutation budgets. Controlled nuisance simulations further confirmed that structurally coherent but target-independent variation can be selected, emphasizing that GSR-FS estimates internal structural relevance rather than guaranteed target-specific relevance. These results support GSR-FS as a reproducible, structurally motivated alternative rather than a statistically superior replacement.
