Nuclear \(\beta\) -decay, a typical decay process for unstable nuclei, is a key mechanism for producing heavy elements in the Universe. In this study, neural networks were employed to predict \(\beta\) -decay half-lives and, for the first time, to identify abnormal trends in nuclear \(\beta\) -decay half-lives based on deviations between experimental values and the predictions of neural networks. Nuclei exhibiting anomalous increases, abrupt peaks, sharp decreases, abnormal odd-even oscillations, and excessively large experimental errors in their \(\beta\) -decay half-lives, which deviate from systematic patterns, were identified through deviations. These anomalous phenomena may be associated with shell effects, shape coexistence, or discrepancies in the experimental data. The discovery and analysis of these abnormal nuclei provide a valuable reference for further investigations using sophisticated microscopic theories, potentially offering insights into new physics through studies of nuclear \(\beta\) -decay half-lives.
