Conference

This paper investigates how image size and preprocessing impact unsupervised clustering, introducing a late-fusion size injection strategy that significantly improves performance on heterogeneous real-world datasets.

This paper introduces a novel ’network surgery’ approach that integrates CNNs with CLIP to provide caption-based explanations. By moving beyond potentially misleading saliency maps, this method identifies the dominant semantic concepts driving a model’s prediction, enabling better detection of spurious biases and improving overall model robustness.

Trade-that! is an algorithmic trading framework designed to classify market states rather than predict raw price movements. By generating target labels from future data and selecting features via a custom ’label separation power’ metric, the engine trains multiple independent classifiers. These are combined into a robust ensemble strategy that optimizes position sizes and adapts to market dynamics, demonstrating significant total returns across diverse market trends.

The “Institut für Werkstofftechnik und Kunststoffverarbeitung” (IWK) is a leading Swiss institute in the area of materials …