Ladder Capsule Network

Authors: Taewon Jeong, Youngmin Lee, Heeyoung Kim

ICML 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experiments on MNIST demonstrate that the ladder capsule network learns an equivariant representation and improves the capability to extrapolate or generalize to pose variations.
Researcher Affiliation Academia 1Department of Industrial and Systems Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea. Correspondence to: Heeyoung Kim <heeyoungkim@kaist.ac.kr>.
Pseudocode Yes Algorithm 1 Dynamic routing algorithm (Sabour et al., 2017)
Open Source Code Yes The code for L-Caps Net is available at https://github.com/taewonjeong/L-Caps Net.
Open Datasets Yes The experiments on MNIST demonstrate that the ladder capsule network learns an equivariant representation and improves the capability to extrapolate or generalize to pose variations.
Dataset Splits No The paper specifies training and testing set sizes (60,000 training, 10,000 testing for MNIST), but does not explicitly mention a separate validation set or specify train/validation/test split percentages beyond a training and testing split.
Hardware Specification No The paper does not explicitly state the specific hardware (e.g., GPU/CPU models, memory) used to run the experiments. It only mentions 'average computation time' without hardware context.
Software Dependencies No The paper mentions software components like 'Adam optimizer' and 'Re Lu activation function' but does not specify any software libraries or dependencies with their version numbers.
Experiment Setup Yes We trained the L-Caps Net with the margin loss with m+ = 0.9, m = 0.1, and λ = 0.5. In addition, we found that adding the loss of difference between the code vector and lower-level activity level, cl Al 2, would be helpful for training; thus we trained the L-Caps Net with the loss L = Lmargin + ϵ cl Al 2 with ϵ = 0.0001. We used the Adam optimizer with exponentially decaying learning rate starting from 0.001.