RI: Medium: Machine Learning for Agricultural and Medical Entomology

这一奖项将使加州大学河滨分校的一个由计算机科学家和昆虫学家组成的团队能够开发传感器和软件，从而对飞行昆虫进行分类。自动而准确地对飞行昆虫进行分类的能力可能会对人类事务产生重大影响，因为昆虫传播疾病，以农作物和牲畜为食，破坏食品店，每年的总成本高达数十亿美元，人类遭受的痛苦也难以估量。该项目的智力价值在于，它可以产生算法、设备和程序，从根本上扩展进行昆虫监测的能力。传感器技术和机器学习的最新进展，以及正在进行的大数据革命，才刚刚开始推动先进算法的开发，这将有助于开创计算昆虫学的新时代。研究人员将建造廉价的设备，可以检测和分类飞行的昆虫。对于至少一些昆虫属，所得到的分类标签将超越物种水平，以预测单个昆虫的性别和生理状态(如处女与怀孕，新出现与成熟)。研究人员将创建计算设备，可以确定选定昆虫的来源，并选择性地捕获目标昆虫，用于下游分子诊断分析。提供这样的信息将加速昆虫学的基础研究，并将允许更有效的病媒控制。该项目的更广泛影响在于有可能显著提高昆虫监测的质量和数量，从而能够更有效地进行综合病媒管理。在蚊子的情况下，已知更有效的干预措施可以直接拯救生命。由此产生的算法将允许创建在多个范围内提供可操作信息的系统，从通知政策委员会到指示农业机器人打开阀门。这些项目全面的教育和外联活动已经在小规模试行，其中包括在K-12和大学一级接触服务不足的社区的详细计划。

This award will enable a team of computer scientists and entomologists at the University of California-Riverside to develop sensors and software that will allow the classification of flying insects. The ability to automatically and accurately classify flying insects has the potential to have significant impact human affairs, because insects spread disease, feed on crops and livestock, and ruin food stores, at a combined annual cost of billions of dollars and incalculable human suffering. The intellectual merit of the project is in producing algorithms, devices, and procedures that will radically expand the ability to conduct insect surveillance. Recent advances in sensor technology and machine learning and the ongoing revolution in Big Data are just beginning to enable development of advanced algorithms that will help usher in a new era of computational entomology. The investigators will build inexpensive devices that can detect and classify flying insects. For at least some genera of insects the resulting classification labels will go beyond species-level to predict sex and physiological states (such as virgin vs. gravid and newly emerged vs. mature) of individual insects. The investigators will create computational devices that can determine the origin of selected insects, and selectively capture targeted insects for downstream molecular diagnostic analysis. Producing such information will both accelerate basic research in entomology and will allow more effective vector control. The broader impacts of the project are inherent in the potential to significantly improve the quality and volume of insect surveillance, thus allowing more effective Integrated Vector Management. In the case of mosquitoes, more effective interventions are known to directly save lives. Resulting algorithms will allow the creation of systems to provide actionable information on multiple scales, from informing a policy committee to instructing an agricultural robot to open a valve. The projects comprehensive educational and outreach activities have already been piloted on a small scale and include detailed plans to reach out to underserved communities at the K-12 and college levels.