مطالب

Out-of-domain intent detection in natural language understanding systems faces significant challenges from suboptimal threshold selection and signal degradation through inappropriate normalization techniques. This paper presents an adaptive ensemble thresholding framework that substantially extends our previous conference work by addressing fundamental limitations in existing variational autoencoder-based detection methods. Our approach combines reconstruction loss from variational autoencoders with classifier confidence scores to create a unified detection signal that captures both semantic deviation and prediction uncertainty. The framework incorporates a novel smart scaling strategy that preserves natural separation ratios between in-domain and out-of-domain samples, preventing the signal destruction caused by standard normalization approaches. Through systematic parameter optimization using grid search techniques, the method adaptively determines optimal ensemble weights and threshold selection strategies tailored to specific dataset characteristics. We evaluate our framework across multiple datasets with varying semantic complexity and domain structures, demonstrating consistent performance improvements over baseline variational autoencoder approaches and recent state-of-the-art methods. Compared to our previous VAE-based approach, the framework demonstrates an average performance gain of 3.15 percentage points across all evaluation metrics. Our analysis reveals that ensemble scaling strategy significantly impacts detection performance, with proper signal preservation being more critical than sophisticated threshold selection methods. This work provides a principled approach to adaptive ensemble learning for out-of-domain detection, offering a robust solution that generalizes effectively across diverse datasets and linguistic contexts including low-resource languages like Persian.

Background: Telecommunications networks are exposed to numerous issues concerning equipment and that causes network outage, which proves very expensive. Basic maintenance methodologies like reactive or even scheduled preventive maintenance cannot cope up with the increasing trends in the facilities of telecom companies. Objective: The article examines how AI is applied to support predictive maintenance so that telecommunication networks can perform as intended with reduced downtime. Methods: The review of existing AI algorithms is presented, focusing on the ML models and deep learning methods. Network operations and maintenance logs are analyzed for data to assess the capabilities of the AI models in terms of prediction. It identifies and analyses such quantifiable parameters as the failure rate prediction accuracy and the response time cut. Results: Computerisation of the forecast maintenance revealed a corresponding decrease in equipment failure incidences and generally reduced time lost due to unscheduled stops. Through the improved network performance, the response to potential threats was quicker than before and services became more reliable and inexpensive to offer. Conclusion: To reduce network outages, reduce network vulnerability, and maximize the efficiency of telecommunications operations, the use of AI-based predictive maintenance can be viewed as a prospect. As technology advances, newer versions of AI algorithms will provide improved predictive strength and incorporation into the telecommunications system.