Machine learning on GPUs and application to data science

GPU 上的机器学习及其在数据科学中的应用

• 批准号: 2227382
• 金额: --
• 依托单位: Swansea University
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2019
• 资助国家: 英国
• 起止时间: 2019 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2227382
• 关键词: Machine; learning; GPUs; application; data

When I followed a computer vision module at Swansea University, I realized that Machine Learning is a field in which I am interested.Machine learning have shown high potential in terms of providing customized services in Healthcare [1], Financial Services [2] and etc. While machine learning can be used for the greater good, it also can be dangerous if there is not enough control on it. For instance, personal data from millions people were harvested by Cambridge Analytica [3] and used for political purposes, which presents ethical issues. At Helping Human Beings, I would like to use AI's potential in medical area such as diagnose disease which could alleviate the burden for the NHS and help RDAG (Rare Diseases Advisory Group).One important need in rare diseases is timely diagnosis which can play a significant life changing rule in life of patients [4]. While it is relatively easy to warn someone about his/her diabetic risks for instance, other diseases may be more difficult to diagnose. Abundance of data in medical care are generated and stored either for administrative purposes or for research [5]. Therefore I can easily foresee that one could get better diagnostics by using this data and this would be of interest to millions of people if data confidentiality and privacy can be respected.As such, I would be passionate about working on any fascinating AI techniques that can support people similarly to the case mentioned previously, without being detrimental to the people.

当我在斯旺西大学学习计算机视觉模块时，我意识到机器学习是我感兴趣的一个领域。机器学习在医疗保健[1]，金融服务[2]等提供定制服务方面显示出很高的潜力。虽然机器学习可以用于更大的利益，但如果没有足够的控制，它也可能是危险的。例如，剑桥分析公司收集了数百万人的个人数据[3]，并将其用于政治目的，这带来了伦理问题。在Helping Human Beings，我想利用AI在医疗领域的潜力，例如诊断疾病，可以减轻NHS的负担，并帮助RDAG（罕见疾病咨询小组）。罕见疾病的一个重要需求是及时诊断，这可以在患者的生活中发挥重要的生命改变规则[4]。虽然它是相对容易警告某人他/她的糖尿病风险，例如，其他疾病可能更难以诊断。医疗保健中的数据摘要是出于管理目的或研究目的而生成和存储的[5]。因此，我可以很容易地预见，人们可以通过使用这些数据来获得更好的诊断，如果数据的机密性和隐私能够得到尊重，这将对数百万人产生兴趣。因此，我将热衷于研究任何有趣的人工智能技术，这些技术可以像前面提到的情况一样支持人们，而不会对人们造成伤害。