Probabilistic Attention-to-Influence Neural Models for Event Sequences

Authors: Xiao Shou, Debarun Bhattacharjya, Tian Gao, Dharmashankar Subramanian, Oktie Hassanzadeh, Kristin Bennett

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We motivate our general framework and show improved performance in experiments compared to existing baselines on synthetic data as well as realworld benchmarks, for tasks involving prediction and influencing set identification.
Researcher Affiliation Collaboration 1Rensselaer Polytechnic Institute, Troy, NY, USA 2IBM AI Research, Yorktown Heights, NY, USA.
Pseudocode Yes Algorithm 1 Topology-based event sequence generator (with Python pseudo code)
Open Source Code No The paper does not contain an explicit statement or link indicating that the authors have released open-source code for their proposed model. It mentions a GitHub link in the appendix related to a baseline (THP), but not their own code.
Open Datasets Yes Datasets. We consider 5 real event datasets in different domains curated previously (Bhattacharjya et al., 2022). ... Diabetes (Frank & Asuncion, 2010) ... Stack Overflow (Grant & Betts, 2013) ... Linked In (Xu et al., 2017) ... Beige Books ... Timelines ... We show an example of influencing set discovery by our model Uniform-τ on a dataset derived from a corpus of news article snippets from Event Registry (Leban et al., 2014).
Dataset Splits Yes Each dataset is randomly split into 70%-15%-15% train, dev, and test set.
Hardware Specification Yes All our experiments are performed on a private server (https://idea.rpi.edu/IDEA Cluster Access) with TITAN RTX GPU.
Software Dependencies No The paper mentions implementing components in PyTorch and using the Adam optimizer, but it does not provide specific version numbers for these software dependencies (e.g., PyTorch 1.x, Adam 2.x).
Experiment Setup Yes The experiment setting for hyperparameters in Uniform-2 and Sparse-2 for binary prediction is given in Table 7. The τ values for Uniform-τ are {0.4,0.5,0.6} and for Sparse-τ they are {0.1,0.2,0.3}.