KNOWLEDGE DISCOVERY FROM ROAD TRAFFIC ACCIDENT DATA IN ETHIOPIA: DATA QUALITY, ENSEMBLING AND TREND ANALYSIS FOR IMPROVING ROAD SAFETY
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of slow convergence and premature convergence in Genetic Algorithms (GAs), which are widely used for search and optimization but often struggle to fine-tune solutions or escape local optima. The authors propose a hybrid algorithm that integrates Artificial Neural Networks (ANNs) into the parent selection process of a GA. The motivation is to move beyond traditional selection mechanisms that rely solely on individual fitness values, which do not guarantee that two high-fitness parents will produce superior offspring. Instead, the proposed method aims to identify structural correlations and coherent building blocks between parents that increase the likelihood of producing high-fitness offspring, thereby accelerating convergence. The methodology involves a two-phase process. Initially, the GA runs with standard tournament selection to collect training data, consisting of parent pairs and the fitness of their resulting offspring. After a predefined number of generations, this data is used to train a three-layer feedforward neural network using the Rprop learning algorithm. The ANN learns to predict whether a specific pair of parents is likely to produce successful offspring. In the subsequent phase, the first parent is selected via tournament selection, while the second parent is chosen by querying the ANN with potential mates; the individual yielding the highest positive prediction from the network is selected. The algorithm was tested on seven benchmark functions (including Rastrigin, Schwefel, Rosenbrock, Shubert, Shekel’s Foxholes, Colville, and Griewank) using binary encoding, a population size of 60, and 50 runs per test. Results were compared against a Standard Genetic Algorithm (SGA) and an Offspring Selection Genetic Algorithm (OSGA). The experimental results demonstrate that the hybrid algorithm (GANN) significantly outperforms both SGA and OSGA in terms of average best fitness and stability. For six of the seven benchmark functions, GANN achieved lower average fitness values and smaller standard deviations than OSGA, indicating more robust and accurate solutions. Statistical t-tests confirmed that the improvements were statistically significant for all functions. Furthermore, the percentage of successful crossover operations—where offspring surpassed their parents—was drastically higher in GANN (76–79%) compared to SGA (1–5%). Even when accounting for the additional computational time required for ANN training and inference, GANN maintained superior performance on most functions, with statistical significance retained for five of the seven benchmarks. The significance of this work lies in demonstrating that ANNs can effectively learn the complex structural patterns required for successful genetic recombination. By replacing purely fitness-based selection with a learned predictive model, the hybrid approach reduces the randomness of crossover outcomes and enhances the efficiency of the search process. This suggests that integrating neural networks into evolutionary algorithms can provide a powerful mechanism for improving convergence speed and solution quality in optimization problems.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | OpenAlex-citations | — | — | 1 | 2026-06-19 |
| archive | success | semantic_scholar | — | — | 6 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | partial | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: crash risk outcomes
- Methodological Resource: dataset resource