Label-Free Supervision of Neural Networks with Physics and Domain Knowledge
Authors: Russell Stewart, Stefano Ermon
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the effectiveness of this approach on real world and simulated computer vision tasks. In our first two experiments, we construct a mapping from an image to the location of an object it contains. |
| Researcher Affiliation | Academia | Russell Stewart , Stefano Ermon Department of Computer Science, Stanford University {stewartr, ermon}@cs.stanford.edu |
| Pseudocode | No | The paper does not contain explicit pseudocode or algorithm blocks. |
| Open Source Code | No | The paper provides a link for 'Our data set 1' (footnote 1: https://github.com/russell91/labelfree) but does not explicitly state that the code for the methodology is open-source or available at this link. |
| Open Datasets | Yes | Our data set 1 is collected on a laptop webcam running at 10 frames per second (Δt = 0.1s). [footnote 1: https://github.com/russell91/labelfree] |
| Dataset Splits | No | The paper mentions holding out 25 trajectories for evaluation, which is later referred to as the 'test images'. It does not explicitly specify a separate validation set or provide comprehensive train/validation/test dataset splits with percentages or counts for all three. |
| Hardware Specification | No | The paper mentions data collection on a 'laptop webcam' but does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for training or running the experiments. |
| Software Dependencies | No | The paper mentions using 'Tensor Flow' but does not specify a version number or other software dependencies with their versions. |
| Experiment Setup | Yes | Images are resized to 56 56 pixels... We use 3 Conv/Re LU/Max Pool blocks followed by 2 Fully Connected/Re LU layers with dropout probability 0.5 and a single regression output. We group trajectories into batches of size 16... We use the Adam optimizer... with a learning rate of 0.0001 and train for 4,000 iterations. |