KDD 2025 - Noise-Resilient Point-wise Anomaly Detection in Time Series Using Weak Segment Labels