Department of Information and Communications Engineering, Tokyo Institute of Technology, Tokyo, Japan.
Laboratory for Future Interdisciplinary Research in Science and Technology, Tokyo Institute of Technology, Tokyo, Japan.
Sci Rep. 2022 Sep 29;12(1):16297. doi: 10.1038/s41598-022-20388-0.
Most of the olfactory perception works focused on forward prediction of odor impression, for example, given an odorant's molecular structure parameters or the sensing data predict its odor impression. So far, mapping of mass spectrum of odorant molecules into the odor perception space (binary or continuous sensory space) has been successfully performed. However, it is difficult to predict odorant's sensing data associated with binary odor descriptors (e.g., minty, peach, vanilla etc.). In this study, we have proposed a method to extract the corresponding sensing data (mass spectrum as sensing data) for a desired scent impression although one-to-one relationships are not usually guaranteed. Our target is to extract the sensing data for a given odor descriptor that will help perfumers to create scent. This study is first report for predicting sensing data for a given binary odor descriptor.
大多数嗅觉感知研究都集中在前馈预测气味印象上，例如，给定一种气味物质的分子结构参数或传感数据，预测其气味印象。到目前为止，已经成功地将气味物质的质谱映射到嗅觉感知空间（二进制或连续感官空间）中。然而，很难预测与二元气味描述符（例如薄荷、桃、香草等）相关的气味物质的传感数据。在这项研究中，我们提出了一种方法，即使不能保证一一对应关系，也可以提取出与所需气味印象相对应的传感数据（作为传感数据的质谱）。我们的目标是提取给定气味描述符的传感数据，这将有助于调香师创造香味。这是首次报道预测给定二元气味描述符的传感数据。
