基本信息
浏览量:116
职业迁徙
个人简介
Current Research:
- Novel approaches to automated cyber defence, e.g. development of the first peer reviewed approach to machine-learning based prediction of cyber attack behaviour, within seconds of its execution. Subsequently, how to capture and kill the computational processes behind the cyber attack. In the case of Ransomware, our method reduced file encryption by 92%.
- Potential harms of AI - e.g. if deployed without thorough verification or long-term management. Examples include understanding how attackers can manipulate their behaviour to evade detection by AI-based cyber defence techniques, and how to improve the resilience of AI-driven cyber defences. We also demonstrated that machine learning models can actually be using "hidden" features within the data used to train them to make decisions - such as malware being detected by its evasive behaviour rather than its actual malicious payload.
- Socio-technical security and the understanding of risks to society from cyber-enabled systems. I work very closely with the Criminology team at the Cardiff School of Social Sciences. We have pioneered innovation in AI for the modelling and understanding of risks to social safety and security in online social networks (e.g. production and propagation of cyber hate speech, suicidal ideation and malicious Web links). We have developed methods to understand the emotional factors contributing to propagation of harmful content, as well as network modeling methods to support the management of online harms. This research is organised under the banner of the Social Data Science Lab and HateLab within which I am a director and the computational lead. The Labs' core funding comes from ESRC and forms part of the £64m ‘Big Data Network’
- Novel approaches to automated cyber defence, e.g. development of the first peer reviewed approach to machine-learning based prediction of cyber attack behaviour, within seconds of its execution. Subsequently, how to capture and kill the computational processes behind the cyber attack. In the case of Ransomware, our method reduced file encryption by 92%.
- Potential harms of AI - e.g. if deployed without thorough verification or long-term management. Examples include understanding how attackers can manipulate their behaviour to evade detection by AI-based cyber defence techniques, and how to improve the resilience of AI-driven cyber defences. We also demonstrated that machine learning models can actually be using "hidden" features within the data used to train them to make decisions - such as malware being detected by its evasive behaviour rather than its actual malicious payload.
- Socio-technical security and the understanding of risks to society from cyber-enabled systems. I work very closely with the Criminology team at the Cardiff School of Social Sciences. We have pioneered innovation in AI for the modelling and understanding of risks to social safety and security in online social networks (e.g. production and propagation of cyber hate speech, suicidal ideation and malicious Web links). We have developed methods to understand the emotional factors contributing to propagation of harmful content, as well as network modeling methods to support the management of online harms. This research is organised under the banner of the Social Data Science Lab and HateLab within which I am a director and the computational lead. The Labs' core funding comes from ESRC and forms part of the £64m ‘Big Data Network’
研究兴趣
论文共 193 篇作者统计合作学者相似作者
按年份排序按引用量排序主题筛选期刊级别筛选合作者筛选合作机构筛选
时间
引用量
主题
期刊级别
合作者
合作机构
Big Data and Cognitive Computingno. 4 (2024)
CoRR (2024)
引用0浏览0EI引用
0
0
2023 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)pp.282-284, (2023)
引用0浏览0EIWOS引用
0
0
EuroS&P Workshopspp.370-378, (2023)
引用0浏览0EIWOS引用
0
0
CoRR (2023)
引用0浏览0EI引用
0
0
加载更多
作者统计
合作学者
合作机构
D-Core
- 合作者
- 学生
- 导师
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn