Auto-Encoding Sequential Monte Carlo
Authors: Tuan Anh Le, Maximilian Igl, Tom Rainforth, Tom Jin, Frank Wood
ICLR 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 5 EXPERIMENTS We now present a series of experiments designed to answer the following questions: 1) Does tightening the bound by using either more particles or a better inference procedure lead to an adverse effect on proposal learning? 2) Can AESMC, despite this effect, outperform IWAE? 3) Can we further improve the learned model and proposal by using ALT? |
| Researcher Affiliation | Academia | Department of Engineering Science, University of Oxford Department of Statistics, University of Oxford Department of Statistics, University of Warwick |
| Pseudocode | Yes | Algorithm 1: Sequential Monte Carlo |
| Open Source Code | No | The paper does not contain any statement about making its source code publicly available or provide a link to a code repository. |
| Open Datasets | No | The dataset is inspired by (Ondrúška & Posner, 2016), however with the crucial difference that the movement of the agent is stochastic. The paper describes the dataset but does not provide a direct link, DOI, specific repository, or a clear statement of public availability for the modified dataset used in the experiments. |
| Dataset Splits | No | The paper states '1000 are randomly held out as test set' for the Moving Agents dataset, but it does not provide explicit details about training/validation/test splits, specific percentages, or how a validation set was used for hyperparameter tuning. |
| Hardware Specification | No | The paper does not provide any specific details regarding the hardware used to run the experiments, such as GPU models, CPU types, or memory. |
| Software Dependencies | No | The paper mentions software components like 'ADAM' and 'Gated Recurrent Unit (GRU)' and describes network architectures, but it does not provide specific version numbers for any software dependencies or libraries. |
| Experiment Setup | Yes | Architecture and hyperparameter details are given in Appendix C.1. For the generative model and proposal distribution we use a Variational Recurrent Neural Network (VRNN) (Chung et al., 2015). The functions µx θ and σx θ are computed by networks with two fully connected layers of size 128 whose first layer is shared. For the moving agents dataset we use ADAM with a learning rate of 10-3. We use a minibatch size of 25. |