Notice: This website is continuously updated as new conference proceedings and journal volumes are published. Consequently, the figures presented here may differ from those reported in our publications.
A brief overview of our methodology is provided on the Methods page, along with references to publications documenting our approach and links to the code and data used in each study.
The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates.
Doubly Stochastic Variational Inference for Neural Processes with Hierarchical Latent Variables
Authors: Qi Wang, Herke Van Hoof
ICML 2020 | Venue PDF | LLM Run Details | Input Tokens: 19,453 Total number of tokens sent to the LLM as input for this paper's analysis. | Output Tokens: 5,498 Total number of tokens produced by the LLM (including reasoning/thinking tokens) for this paper's analysis.
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate this model in several experiments, and our results demonstrate competitive prediction performance in multi-output regression and uncertainty estimation in classification. In this section, we start with learning predictive functions on several toy dataset, and then high-dimensional tasks, including system identification on physics engines, multioutput regression on real-world dataset as well as image classification with uncertainty quantification, are performed to evaluate properties of NP related models. |
| Researcher Affiliation | Academia | 1Amsterdam Machine Learning Lab, University of Amsterdam, Amsterdam, the Netherlands. Correspondence to: Qi Wang <EMAIL>. |
| Pseudocode | Yes | Algorithm 1 Variational Inference for DSVNP in Training. |
| Open Source Code | No | The paper does not provide an explicit statement or link for open-source code for the described methodology. |
| Open Datasets | Yes | We evaluate the performance of all models on dataset, including SARCOS 2, Water Quality (WQ) (Dˇzeroski et al., 2000) and SCM20d (Spyromitros Xioufis et al., 2016). Details about these dataset and neural architectures for all models are included in Appendix E. We respectively train models on MNIST and CIFAR10. |
| Dataset Splits | No | The paper mentions splitting datasets into training and testing sets (e.g., 'Each dataset is randomly split into 2-folds as training and testing sets.'), but does not provide specific details on a separate validation set split or its size/proportion. |
| Hardware Specification | Yes | Also, we gratefully acknowledge the support of NVIDIA Corporation with the donation of a Titan V GPU. |
| Software Dependencies | No | The paper mentions implementation details are in Appendix E, but within the provided text, it does not specify concrete software dependencies with version numbers (e.g., library or solver names with version numbers). |
| Experiment Setup | Yes | The training process follows Algorithm (1) with the maximum number of context points as 100. The dimensions for latent variables are 64 on MNIST and 128 on CIFAR10. The training procedure in (C)NPs follows that in Algorithm (1), and some context points are randomly selected in batch samples. |