Towards Learning Geometric Eigen-Lengths Crucial for Fitting Tasks
Authors: Yijia Weng, Kaichun Mo, Ruoxi Shi, Yanchao Yang, Leonidas Guibas
ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We explore potential solutions and demonstrate the feasibility of learning eigen-lengths from simply observing successful and failed fitting trials. We also attempt geometric grounding for more accurate eigen-length measurement and study the reusability of the learned eigen-lengths across multiple tasks. |
| Researcher Affiliation | Collaboration | 1Stanford University 2NVIDIA Research 3Shanghai Jiaotong University 4HKU. |
| Pseudocode | No | The paper describes network architectures and processes, but does not include any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper mentions a 'Project page: https://yijiaweng.github.io/geoeigen-length', but this is typically a demonstration or information page and not explicitly stated to host the source code for the methodology described in the paper. |
| Open Datasets | Yes | For objects to be fitted in tasks (a)-(e), we use 1200 common household object models from 8 training and 4 testing categories in Shape Net (Chang et al., 2015), following Mo et al. (2021b). |
| Dataset Splits | No | The paper states 'we generated 75k training and 20k testing environment-object pairs' but does not explicitly mention a separate validation split or its size. |
| Hardware Specification | Yes | All experiments are run on a single NVIDIA TITAN X GPU. |
| Software Dependencies | No | The paper states 'All networks are implemented using Py Torch', but it does not provide a specific version number for PyTorch or any other software dependency. |
| Experiment Setup | Yes | All networks are implemented using Py Torch and optimized by the Adam optimizer, with a learning rate starting at 10 3 and decay by half every 10 epochs. Each batch contains 32 data points; each epoch contains around 1600 batches. We train models for 100 epochs on all tasks. |