A Multi-Channel Neural Graphical Event Model with Negative Evidence
Authors: Tian Gao, Dharmashankar Subramanian, Karthikeyan Shanmugam, Debarun Bhattacharjya, Nicholas Mattei3946-3953
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate our method against state-of-the-art baselines on model fitting tasks as gauged by log-likelihood. Through experiments on both synthetic and real-world datasets, we find that our proposed approach outperforms existing baselines on most of the datasets studied. |
| Researcher Affiliation | Collaboration | Research AI, IBM T. J. Watson Research Center, Yorktown Heights, NY, USA Department of Computer Science, Tulane University, New Orleans, LA, USA {tgao, dharmash, debarunb}@us.ibm.com Karthikeyan.Shanmugam2@ibm.com, nsmattei@tulane.edu |
| Pseudocode | No | The paper describes the model architecture and training process in text and diagrams, but does not include explicit pseudocode or algorithm blocks. |
| Open Source Code | No | We implement MCN-GEM in Pytorch, PGEM in Python, and use publicly available code for NHP1. 1https://github.com/HMEIat JHU/neurawkes |
| Open Datasets | Yes | We use the Integrated Crisis and Early Warning System (ICEWS) political event dataset (O Brien 2010) to test our model. ... Following Mei and Eisner (2017), we also test algorithms on the MIMIC-II dataset... The data is generated following the same sampling procedure described in Bhattacharjya, Subramanian, and Gao (2018). |
| Dataset Splits | No | For our experiments, we divide an event dataset into 70% 30% train-test splits. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for running the experiments. |
| Software Dependencies | No | We implement MCN-GEM in Pytorch, PGEM in Python, and use publicly available code for NHP1. |
| Experiment Setup | Yes | Note that K is a hyperparameter, and Figure 2(b) shows one fake epoch (K = 1) introduced at ti,i+1. ... Note that J is a constant hyperparameter in the model. ... Ltrain = LL(D) + λp Lp + λw Lw (3) ... We study the impact of the number of fake epochs on the performance of MCN-GEM through a study on the ICEWS Argentina dataset. We vary the number of fake epochs, from 0 to 5... memory bank sizes (M=1,2,3,5,10). |