Maximum principle for partially observed risk-sensitive optimal control problems of mean-field type · Study