Fixing Implicit Derivatives: Trust-Region Based Learning of Continuous Energy Functions
Authors: Chris Russell, Matteo Toso, Neill Campbell
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show how our approach can be applied to obtain state-of-the-art results in the diverse applications of tracker fusion and multiview 3D reconstruction. |
| Researcher Affiliation | Academia | Matteo Toso CVSSP, University of Surrey Neill D. F. Campbell University of Bath Chris Russell CVSSP, University of Surrey and The Alan Turing Institute |
| Pseudocode | No | The paper does not contain clearly labeled 'Pseudocode' or 'Algorithm' blocks. Figure 1 contains mathematical function definitions, not pseudocode. |
| Open Source Code | Yes | Code is available at: https://github.com/MatteoT90/WibergianLearning |
| Open Datasets | Yes | We evaluate our approach on the Human3.6M dataset[40], over the standard training and testing sets for both monocular and multiview reconstruction. |
| Dataset Splits | Yes | For evaluation purposes we divide the VOT2018 dataset roughly into two thirds training/one third test by number of frames, so that no sequence occurs in both halves, training on the first half of the frames and reporting loss on the second; we also report on missing at random frames , where frames from the same sequence occur in both training and test. |
| Hardware Specification | No | The paper mentions 'a typical CPU' and '120 core hours' but does not specify any particular CPU model (e.g., Intel i7, Xeon), GPU models, or other specific hardware details used for running experiments. |
| Software Dependencies | No | The paper mentions general software components like L-BFGS but does not provide specific version numbers for any libraries, frameworks, or programming languages used (e.g., PyTorch 1.9, Python 3.8). |
| Experiment Setup | Yes | In practice no tuning or online adaption of the value of λ is needed, and we simply set it to 0.1. ... Instead we use stochastic gradient descent with a strong momentum value of 0.999 to damp the oscillations. |