Design and development of machine learning and data optimization processes for detection of biogenic patterns

设计和开发用于检测生物模式的机器学习和数据优化流程

• 批准号: 490833-2015
• 负责人: Zamar, Ruben
• 金额: $ 1.82万
• 依托单位: University of British Columbia
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2015
• 资助国家: 加拿大
• 起止时间: 2015-01-01 至 2016-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=586642
• 关键词: Design; development; machine; learning; data

EIS is working on the development of a wireless plant health monitoring system that is capable of gathering information from crops through a series of sensory systems that monitor and interpret different plant signal patterns. This approach constitutes of three main steps: (1) Identifying and refining plant-driven measurable indicators; (2) designing, and customizing high-resolution sensory systems that is deployable in various agricultural environment (e.g., greenhouses and vineyards); (3) developing interpretation algorithms and reporting software; and (4) designing wireless machine-to-machine (M2M) solutions for non-invasive implementation of the technology. The underlying challenges of such technology are the infrastructure to handle large amount of data, advanced processing and data analytics and safe keeping and encryption. The main focus of this project will be on the third step. In collaboration with EIS who has already developed some interpretation algorithms and models, we will review and assess the nature of the sensory data and develop novel data optimization and machine learning processes. The objectives are to (i) develop a data reduction pipeline to obtain an optimal and interpretable subset of the sensory data suitable for predicting plant infestation; (ii) develop an algorithm for constructing a predictive model that uses sensory information to assign a given plant in an agricultural environment an infection status; (iii) should time permit, design a set of processes that can transfer information from a given agricultural environment to establish a classification model for another agricultural environment for real time use.

EIS正在开发一种无线工厂健康监测系统，该系统能够收集 通过一系列监测和解释不同植物信号的感觉系统从作物中获取信息 模式.该方法由三个主要步骤组成：（1）识别和提炼工厂驱动的可测量 指标;（2）设计和定制可在各种环境中部署的高分辨率传感系统 农业环境（例如，温室和葡萄园）;（3）开发解释算法， 报告软件;以及（4）设计无线机器对机器（M2M）解决方案， 技术的实施。此类技术的潜在挑战是基础设施， 处理大量数据、高级处理和数据分析以及安全保管和加密。的 该项目的主要重点将放在第三步。与已经开发了一些 解释算法和模型，我们将审查和评估感官数据的性质，并开发 新的数据优化和机器学习过程。目标是：（一）减少数据 管线，以获得适于预测设备的传感数据的最佳且可解释的子集 （ii）开发用于构建预测模型的算法，该预测模型使用感官信息来 指定农业环境中的给定植物的感染状态;（iii）如果时间允许，设计一套 可以从给定的农业环境中转移信息以建立分类的过程 用于另一农业环境的模型，用于真实的时间使用。