TY - GEN
T1 - Psychological measure on fish catches and its application to optimization criterion for machine-learning-based predictors
AU - Kokaki, Yuya
AU - Kobayashi, Tetsunori
AU - Ogawa, Tetsuji
N1 - Funding Information:
Psychological fish catch scaling was proposed to incorporate fishery workers’ intuition in an automatic FCP. Questionnaire surveys on the prediction error tolerance indicated that fishery workers tended to evaluate errors strictly for small catches. In addition, the Weber–Fechner law was refined to derive the psychological fish catch. The training of GBDTs in a psychological error minimization manner was effective in reducing prediction errors, especially when the fish catches were small (e.g., for the end period of fishing). This result implied that the proposed psychological measure could contribute to forecasts suitable for fishermen’s sense. ACKNOWLEDGMENTS This research and development work was supported by the MIC/SCOPE #172302010. The authors would like to thank Ohtomo Suisan Co., Ltd. for sharing the fish catch data, and the project members for the helpful discussions. REFERENCES
Publisher Copyright:
© 2019 IEEE.
PY - 2019/6
Y1 - 2019/6
N2 - Psychological fish catches are designed and successfully applied to an optimization criterion for machine-learning-based predictors. In automatic fish catch prediction, prediction errors allowed by fishery workers differ depending on fish catches. Such error tolerance has not been considered both in evaluating prediction results and in training predictors. In the present study, the investigation of fishermen's tolerance in prediction errors using a psychophysics method and subsequently reflecting it in constructing a psychological measure on fish catches is performed. In addition, the psychological measure obtained is exploited in the optimization of fish catch prediction models to provide forecasts intuitive to fishery workers.
AB - Psychological fish catches are designed and successfully applied to an optimization criterion for machine-learning-based predictors. In automatic fish catch prediction, prediction errors allowed by fishery workers differ depending on fish catches. Such error tolerance has not been considered both in evaluating prediction results and in training predictors. In the present study, the investigation of fishermen's tolerance in prediction errors using a psychophysics method and subsequently reflecting it in constructing a psychological measure on fish catches is performed. In addition, the psychological measure obtained is exploited in the optimization of fish catch prediction models to provide forecasts intuitive to fishery workers.
KW - Constant method
KW - Fish catch prediction
KW - Machine learning
KW - Psychological measure
KW - Psychophysics
UR - http://www.scopus.com/inward/record.url?scp=85103688562&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85103688562&partnerID=8YFLogxK
U2 - 10.1109/OCEANSE.2019.8867405
DO - 10.1109/OCEANSE.2019.8867405
M3 - Conference contribution
AN - SCOPUS:85103688562
T3 - OCEANS 2019 - Marseille, OCEANS Marseille 2019
BT - OCEANS 2019 - Marseille, OCEANS Marseille 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2019 OCEANS - Marseille, OCEANS Marseille 2019
Y2 - 17 June 2019 through 20 June 2019
ER -