Distribution-Interpolation Trade off in Generative Models

Authors: Damian Leśniak, Igor Sieradzki, Igor Podolak

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

Reproducibility Variable Result LLM Response
Research Type Experimental The experiments conducted using the DCGAN model on the Celeb A dataset are presented solely to illustrate the problem, not to study the DCGAN itself, theoretically or empirically. All experiments were conducted using a DCGAN model (Radford et al., 2015), in which the generator network consisted of a linear layer with 8192 neurons, followed by four convolution transposition layers, each using 5 5 filters and strides of 2, with number of filters in order of layers: 256, 128, 64, 3.
Researcher Affiliation Academia Damian Le sniak Jagiellonian University Igor Sieradzki Jagiellonian University Igor Podolak Jagiellonian University
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement or link for open-source code for the methodology described.
Open Datasets Yes The experiments conducted using the DCGAN model on the Celeb A dataset (Liu et al., 2015)
Dataset Splits No The paper mentions using the Celeb A dataset and specific training parameters like batch size, but does not specify the training, validation, or test dataset splits.
Hardware Specification No The paper describes the model architecture and training parameters, but it does not specify any hardware details (e.g., GPU models, CPU types, or memory) used for running the experiments.
Software Dependencies No The paper mentions using the 'Adam optimiser' and 'No batch normalisation' but does not specify any version numbers for software dependencies such as libraries or frameworks.
Experiment Setup Yes Adam optimiser with learning rate of 2e 4 and momentum set to 0.5 was used. Batch size 64 was used throughout all experiments. If not explicitly stated otherwise, latent space dimension was set to 100. For the Celeb A dataset we resized the input images to 64 64.