The field of machine learning is moving towards a greater emphasis on uncertainty quantification and conformal prediction. This is driven by the need for more reliable and robust models that can provide accurate predictions and uncertainty estimates in real-world applications. Recent developments have focused on improving the efficiency and accuracy of conformal prediction methods, as well as their integration with deep learning models. Notable advancements include the development of learnable conformal prediction methods that can adapt to different contexts and tasks, and the application of uncertainty quantification techniques to improve the reliability of models in areas such as soil science and autonomous driving. Noteworthy papers include:

• Learnable Conformal Prediction with Context-Aware Nonconformity Functions for Robotic Planning and Perception, which presents a novel approach to conformal prediction that achieves state-of-the-art results in several robotic tasks.
• A study of Universal ODE approaches to predicting soil organic carbon, which explores the use of Universal Differential Equations for predicting soil organic carbon dynamics and highlights the potential and limitations of this approach.
• Uncertainty-Guided Expert-AI Collaboration for Efficient Soil Horizon Annotation, which demonstrates the effectiveness of conformal prediction in improving the efficiency of human-machine collaboration in soil horizon annotation tasks.
• Calibrating the Full Predictive Class Distribution of 3D Object Detectors for Autonomous Driving, which proposes a method for calibrating the full predictive class distribution of 3D object detectors and evaluates its effectiveness in improving the accuracy and reliability of autonomous driving systems.