A Variational Point Process Model for Social Event Sequences

Authors: Zhen Pan, Zhenya Huang, Defu Lian, Enhong Chen173-180

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on real-world datasets prove effectiveness of our proposed model.
Researcher Affiliation Academia 1Anhui Province Key Laboratory of Big Data Analysis and Application, School of Computer Science and Technology, University of Science and Technology of China {pzhen, huangzhy}@mail.ustc.edu.cn, {liandefu, cheneh}@ustc.edu.cn
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statements or links indicating that the source code for the described methodology is publicly available.
Open Datasets Yes Retweets Dataset (Zhao et al. 2015) and Meme Track Dataset (Leskovec, Backstrom, and Kleinberg 2009).
Dataset Splits No The paper mentions 'We randomly sampled disjoint train and test sets with 20,000 and 2,000 sequences respectively' for the Retweets dataset and similarly for the Meme Track dataset, and 'We split both datasets into training and test sets containing 70% and 30% of samples respectively.' While it specifies train and test splits, it does not explicitly mention a separate validation set.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory) used for running the experiments.
Software Dependencies No The paper states, 'The models are implemented with Tensor Flow (Abadi et al. 2016)'. While TensorFlow is mentioned, a specific version number is not provided, nor are other software dependencies with versions.
Experiment Setup Yes Numbers of hidden nodes of LSTMs for Retweets and Meme Track datasets are 256 and 64, respectively. Networks are 2-layer MLPs, with Re LU activation after the first layer. Dimension of the latent code is 256. Event decoder is a 3-layer MLP... The models are implemented with Tensor Flow (Abadi et al. 2016) and are trained using the Adam (Kingma and Ba 2015) optimizer for 1,000 epochs with batch size 32 and learning rate 0.001.