Reciprocal Learning
Authors: Julian Rodemann, Christoph Jansen, Georg Schollmeyer
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate that numerous machine learning algorithms are specific instances of one single paradigm: reciprocal learning. These instances range from active learning over multi-armed bandits to self-training. We show that all these algorithms not only learn parameters from data but also vice versa: They iteratively alter training data in a way that depends on the current model fit. We introduce reciprocal learning as a generalization of these algorithms using the language of decision theory. This allows us to study under what conditions they converge. The key is to guarantee that reciprocal learning contracts such that the Banach fixed-point theorem applies. In this way, we find that reciprocal learning converges at linear rates to an approximately optimal model under some assumptions on the loss function, if their predictions are probabilistic and the sample adaption is both non-greedy and either randomized or regularized. We interpret these findings and provide corollaries that relate them to active learning, self-training, and bandits. |
| Researcher Affiliation | Academia | Julian Rodemann Department of Statistics LMU Munich j.rodemann@lmu.de Christoph Jansen Computing & Communications Lancaster University Leipzig c.jansen@lancaster.ac.uk Georg Schollmeyer Department of Statistics LMU Munich g.schollmeyer@lmu.de |
| Pseudocode | Yes | Algorithm 1: Incremental Self-Training in Semi-Supervised learning and Algorithm 2: Amending Self-Training in Semi-Supervised learning are provided in Appendix A.1.1. |
| Open Source Code | Yes | Code to reproduce findings can be found in https://github.com/rodemann/simulations-self-training-reciprocal-learning. |
| Open Datasets | Yes | Specifically, we deploy incremental self-training with soft labels on a real world datasets (banknote data)... Data source: Public UCI Machine Learning Repository [21]. |
| Dataset Splits | No | The paper mentions varying percentages of unlabeled data (90%, 80%, 70%) for the self-training experiments, but it does not provide specific train/validation/test splits (e.g., 80/10/10) for the labeled data or overall dataset that would be needed for reproducibility of data partitioning. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments, such as GPU/CPU models, memory, or specific cloud resources. |
| Software Dependencies | No | The paper mentions using a "generalized additive model" and refers to a GitHub repository for code, but it does not specify software dependencies with version numbers (e.g., Python 3.x, PyTorch 1.x). |
| Experiment Setup | No | The paper states that it uses a "generalized additive model" and different "selection criteria" from the literature. It also specifies the percentage of unlabeled data (90%, 80%, 70%). However, it does not provide specific hyperparameter values (e.g., learning rate, batch size, number of epochs, optimizer settings) for the models or training process itself that would constitute detailed experimental setup. |