SBIR Phase I: Cloud Based Artificial Intelligence for Trend Analysis Using Sensor Data

这个小企业创新研究（SBIR）第一阶段项目的更广泛的影响/商业潜力是解决数据解释问题，这是可穿戴技术涌入引起的最重要和增长最快的问题之一。与所有技术一样，可穿戴设备越来越受欢迎，成本也在不断下降。急于赶上这股技术范式浪潮的企业遇到了一个复杂的问题，即如何以一种准确且对消费者有用的方式解释收集到的数据。在这些企业中，有许多公司的产品以前与电脑或智能手机技术完全无关。因此，他们不具备内部专业知识，不仅可以正确收集这些数据，还可以对其进行分析，以找出可能被认为对识别消费者的行为或状况有用的模式。通过提供一个盒装解决方案，使基于机器学习的数据分析成为普通工程师的可能，我们的项目旨在帮助这些企业跨越这一障碍。这个小企业创新研究（SBIR）第一阶段项目旨在为非数据/计算机科学家提供人工智能和机器学习系统。虽然机器学习和人工智能技术如今被广泛应用于从b谷歌搜索到优步打车的许多应用中，但它们仍然是相当深奥的话题，仅仅是理解就需要很高的学习曲线，更不用说应用了。我们计划通过在现有硬件传感器平台上构建一个高度直观的web UI来解决这个问题，这与之前的竞争对手不同。这使我们能够利用硬件的数据收集和处理一致性，以及我们专有的sdk来确保正确标记和干净的数据。因此，我们将更容易开发基本的数字处理滤波器，并将机器学习技术应用于数据，以解决通用分类问题。

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase 1 project is to address the problem of data interpretation, one of the most important and fastest growing issues caused by the influx of wearable technologies. As with all technology, wearable devices are increasing in popularity and decreasing in cost every day. Businesses rushing to catch this wave of technology paradigm are met with the complex problem of how to interpret the data gathered in a way that is accurate and useful to consumers. Many of these businesses are companies with products that were previously completely unrelated with computer or smartphone technologies. As such, they do not have the in-house expertise to not only correctly gather such data, but then analyze it for patterns that could be deemed useful in identifying behavior or conditions for the consumer. By providing a boxed solution that makes machine learning based data analysis possible for the average engineer, our project is aimed to help these businesses cross that hurdle.This Small Business Innovation Research (SBIR) Phase 1 project seeks to bring Artificial Intelligence and Machine Learning systems for use in the hands of non-data/computer scientists. While machine learning and AI techniques are widely used these days in many applications ranging from Google search to Uber rides, they remain fairly esoteric topics with a high learning curve just to understand, let alone apply. We plan to address this issue differently from previous competitors by building a highly intuitive web UI on top of our existing hardware sensor platform. This allows us to leverage the data gathering and processing consistency of our hardware, along with our proprietary SDKs to ensure properly labelled and clean data. As a result, we will have a much easier time developing basic digital processing filters as well as applying machine learning techniques to the data in order to solve generic classification problems.