剑桥大学PhD position in Privacy and robustness in modern machine learning申请条件要求-申请方
主办方
剑桥大学
PhD直招介绍
About the Project
AI_CDT_DecisionMaking
Details
The goal of this project is to investigate new methods to incorporate privacy and/or robustness in machine learning algorithms. Only recently, privacy and robustness have been studied in theoretical statistics, answering questions such as: What is the best possible performance for a statistical estimator which respects a certain level of privacy, or a certain level of robustness? What are algorithms that achieve near-optimal performance? Many algorithms that have been shown to achieve state-of-the-art performance are impractical to implement, and the problems which have been studied rigorously are quite limited in scope. Even the specific notions of (differential) privacy and (adversarial) robustness are evolving topics in theoretical machine learning, and the precise definitions must be considered carefully when deriving rigorous mathematical results.
This project will focus on studying machine learning algorithms for networked-structured data. Some of the sub-problems to be explored involve learning communities in an unlabeled graph, inference in dynamically changing random graphs, and estimation in settings where the graph and edge weights contain information concerning interactions between agents who seek to jointly solve a statistical inference problem. All of these problems have been studied in some form in recent years *without* privacy or robustness constraints, and the goal will be to determine how such constraints can be incorporated into existing algorithmic frameworks in a computationally feasible manner. Since privacy and robustness both concern stability of an algorithm, a higher-level question is to establish a practically efficient pipeline by which private algorithms can be converted into robust algorithms, and vice versa.
The University actively supports equality, diversity and inclusion and encourages applications from all sections of society. We place major emphasis on the importance of team work and an enjoyable work environment as a foundation for performing internationally leading research. This will allow the student to acquire cutting edge research methodologies in a supportive environment, where they can focus on making the best possible scientific progress.
剑桥大学 PhD position in Privacy and robustness in modern machine learning项目有没有奖学金，是不是全奖Phd招生，下面我们一起看一下【大学名称】Phd的奖学金资助情况
项目资助情况
This is a fully funded AI UKRI CDT 4 year program; Home tuition fees will be provided, along with a tax-free stipend (subject to individual circumstances), set at the UKRI rate (e.g. £19,237 for 2024/25) . The start date is September 2025.
Project based in University of Cambridge
剑桥大学Phd申请条件和要求都有哪些？PhD position in Privacy and robustness in modern machine learning项目是不是全奖？有没有奖学金？下面我们一起看一下剑桥大学申请Phd直招需要具备哪些条件和要求，以及托福、雅思语言成绩要到多少才能申请。
申请要求
A strong background in mathematics, with undergraduate coursework in statistics/probability and optimization theory. Some background in machine learning (theory) would also be desirable.
报名方式
申请链接
联系人
邮箱：aidecisionscdt@manchester.ac.uk
相关资源
扫描二维码
下载申请方APP
下载申请方APP
扫描二维码
下载申请方APP