Abstract
Logic Tensor Networks (LTNs) have been shown to be capable of many artificial intelligence tasks, regression being among them. Earth System Models (ESMs) present large opportunities and a need for computationally fast and inexpensive regression solutions. In an approach to the capabilities of this neurosymbolic framework, dozens of equations were modeled from the Community Land Model. Most LTN regression models were able to reach greater than 0.65 axiom satisfaction and falling below an accuracy of 0.779 using the Root Mean Square Error (RMSE). Having yielded positive results, the Logic Tensor Network framework may provide the speed and efficiency desired in ESMs and further consideration of LTNs is justified.
| Original language | English |
|---|---|
| Title of host publication | 2024 IEEE Opportunity Research Scholars Symposium, ORSS 2024 |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 97-100 |
| Number of pages | 4 |
| ISBN (Electronic) | 9798350390698 |
| DOIs | |
| State | Published - 2024 |
| Event | 2024 IEEE Opportunity Research Scholars Symposium, ORSS 2024 - Atlanta, United States Duration: Apr 15 2024 → Jul 15 2024 |
Publication series
| Name | 2024 IEEE Opportunity Research Scholars Symposium, ORSS 2024 |
|---|
Conference
| Conference | 2024 IEEE Opportunity Research Scholars Symposium, ORSS 2024 |
|---|---|
| Country/Territory | United States |
| City | Atlanta |
| Period | 04/15/24 → 07/15/24 |
Funding
The work was funded by NSF Grant 2420355.
Keywords
- Earth System models
- Logic Tensor Networks
- Machine learning