Evidential Turing Processes
This provides a unified solution for safety-critical applications requiring total uncertainty quantification, though it appears incremental as it combines existing techniques.
The paper tackles the problem of probabilistic classifiers needing reliable predictive uncertainties across multiple aspects, and introduces a method combining Evidential Deep Learning, Neural Processes, and Neural Turing Machines that excels in all three aspects of total calibration on five classification tasks.
A probabilistic classifier with reliable predictive uncertainties i) fits successfully to the target domain data, ii) provides calibrated class probabilities in difficult regions of the target domain (e.g.\ class overlap), and iii) accurately identifies queries coming out of the target domain and rejects them. We introduce an original combination of Evidential Deep Learning, Neural Processes, and Neural Turing Machines capable of providing all three essential properties mentioned above for total uncertainty quantification. We observe our method on five classification tasks to be the only one that can excel all three aspects of total calibration with a single standalone predictor. Our unified solution delivers an implementation-friendly and compute efficient recipe for safety clearance and provides intellectual economy to an investigation of algorithmic roots of epistemic awareness in deep neural nets.