| (英) |
In order to revitalize the recreational fishing industry in rivers, efforts are underway to make fishing tickets available online, and it is expected that fishing results forecasting will be realized in the future.
Fishing performance in rivers is greatly affected by factors such as date, time, climate, and water level, and anglers' fishing performance data must be collected to predict fishing performance.
Under these circumstances, the realization of fishing result prediction using federated learning is being considered.
In this system, each angler learns from his or her own catch data using his or her own smartphone, and only the learning results are provided to the angler to realize a catch forecast based on the catch data of various anglers.
In this paper, we conduct an experimental evaluation of the fishing result prediction system to realize a fishing result prediction system using coalition learning.
In this experiment, we construct an experimental system using a part of real data of actual fishing trips, and investigate the accuracy of the prediction through experiments. |