Pairwise Causality Guided Transformers for Event Sequences

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

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate that our approach outperforms several state-of-the-art models in terms of prediction accuracy by effectively leveraging knowledge about causal pairs. ... We conduct a detailed empirical evaluation demonstrating superior predictive performance as compared to state-of-the-art baselines on synthetic datasets as well as real-world benchmarks.
Researcher Affiliation Collaboration Xiao Shou RPI xshou01@gmail.com Debarun Bhattacharjya IBM Research debarunb@us.ibm.com Tian Gao IBM Research tgao@us.ibm.com Dharmashankar Subramanian IBM Research dharmash@us.ibm.com Oktie Hassanzadeh IBM Research hassanzadeh@us.ibm.com Kristin Bennett RPI bennek@rpi.edu
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any explicit statement or link regarding the availability of open-source code for the described methodology.
Open Datasets Yes Diabetes contains meals, exercise activity, insulin dosage, and changes in blood glucose measurements for a group of 65 diabetes patients [53]. ... [53] Andrew Frank and Arthur Asuncion. UCI machine learning repository, 2010.
Dataset Splits Yes We run experiments on 4 generated synthetic datasets to verify the learning capabilities and validity of our approach. ... each dataset was split into train/dev/test sets (60/20/20)%. ... All datasets were split into train-dev-test sets (60/20/20)% for fair evaluation, where hyper-parameters were chosen using train-dev sets and evaluation was performed on the test sets.
Hardware Specification No The paper does not provide specific hardware details (such as exact GPU/CPU models or processor types) used for running its experiments.
Software Dependencies No The paper mentions using a 'transformer architecture' and a 'flan-t5-xxl(11B)' model but does not provide specific software dependencies with version numbers (e.g., library or framework versions, Python version) needed to replicate the experiment.
Experiment Setup No Further details around implementation and training can be found in the Appendix. ... Further details such as hyper-parameters chosen based on the validation set are discussed in the Appendix.