Can Machines Learn the True Probabilities?

Authors: Jinsook Kim

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We prove under some basic assumptions when machines can learn the true objective probabilities, if any, and when machines cannot learn them. The outline of the proof is as follows. After defining some main concepts, we identify the necessary condition for any machine to learn the true objective probabilities. From this necessary condition, we derive the theorem that learning implies the true guarantee of well-calibration.
Researcher Affiliation Academia 1Underwood International College, Yonsei University, Seoul, Korea. Correspondence to: Jinsook Kim <jki76364@gmail.com>
Pseudocode No The paper is purely theoretical, focusing on mathematical proofs, theorems, and definitions. It does not include any pseudocode or algorithm blocks.
Open Source Code No The paper is theoretical and focuses on mathematical proofs and concepts. There is no mention of any open-source code being released for the methodology described.
Open Datasets No The paper is theoretical and does not conduct experiments involving datasets for training or evaluation. While the term 'data' is used, it refers to conceptual 'actual data' generated by Nature within the theoretical framework, not specific experimental datasets.
Dataset Splits No The paper is theoretical and does not involve empirical experiments. Therefore, there are no mentions of specific dataset splits for training, validation, or testing.
Hardware Specification No The paper is theoretical and does not describe any computational experiments. Therefore, no hardware specifications are mentioned.
Software Dependencies No The paper is theoretical and does not describe any computational experiments. Therefore, no software dependencies with specific version numbers are mentioned.
Experiment Setup No The paper is theoretical and does not involve experimental setup details such as hyperparameters, training configurations, or system-level settings.