![](https://static.wixstatic.com/media/84770f_e9e3ddab1b664dfa8e77b6ff4380add8~mv2.jpg/v1/fill/w_581,h_387,al_c,q_80,usm_0.66_1.00_0.01,enc_auto/84770f_e9e3ddab1b664dfa8e77b6ff4380add8~mv2.jpg)
![DSC07857 (1).jpg](https://static.wixstatic.com/media/3e8f5d_703ea1ef995d465281db1ca43cd20245~mv2.jpg/v1/fill/w_420,h_553,al_c,q_80,usm_0.66_1.00_0.01,enc_auto/DSC07857%20(1).jpg)
Adil Khan
Senior Research Scientist and Professor in Deep Learning
I am a Professor of Machine Learning (ML), Director of Postgraduate Research, and Head of the Machine Learning Research Group at the University of Hull. I am also the CTO at AIDecisions.
* AIDecisions is a small UK-based R&D AI firm, established in the summer of 2024.
We will showcase our first product on LLM-guided Instance level Image Manipulations at BMVC 2024.
Research Mission
Generally robust deep models capable of context-aware self-explainability, Novel guidance mechanisms and optimisation frameworks for Generative AI
Addressing critical challenges in robustness of deep models and their interpretability, since current models often fail when confronted with adversarial perturbations and domain shifts, which limits their reliability in real-world applications.
Generally Robust Models
Dynamic Context-aware Explinability
Dynamic, context-aware explainability refers to the ability of an AI system to tailor its explanations based on the context, including the audience’s background, the nature of the task, the specific input data, and other relevant factors. This is akin to how a human expert would adjust their explanations depending on who they’re talking to and what the situation is. The need for this kind of explainability arises from the fact that different users have different levels of understanding, different perspectives, and different needs. A one-size-fits-all explanation will not be effective for everyone.
Designing novel guidance mechanisms and optimisation frameworks to achieve precise and meaningful manipulations of generative model outputs, thereby enhancing their utility, reliability, and applicability in real-world scenarios. This will drive innovation across industries by providing tools that allow for more precise and reliable generative models. Furthermore, the ability to generate high-quality, context-specific content can democratize access to advanced AI tools, making them more accessible to a broader audience and fostering global inclusion.
Guidance for Generative AI
Featured Works
Visit my google scholar for a comprehensive listing of my publications
![image.png](https://static.wixstatic.com/media/3e8f5d_ff9c0f314d1e4c16bf3a0d610c3c55b9~mv2.png/v1/fill/w_236,h_98,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_ff9c0f314d1e4c16bf3a0d610c3c55b9~mv2.png)
I obtained my Ph.D. degree at Kyung Hee University in 2010, advised by Prof. Tae-seong Kim, where my contributions to the field of ML and artificial intelligence were marked by my innovative use of hierarchical approaches and learning paradigms to solve complex real-life challenges across diverse domains. My pioneering work in physical-activity recognition using triaxial accelerometers and hierarchical recognisers significantly advanced healthcare and eldercare technologies.
![image.png](https://static.wixstatic.com/media/3e8f5d_08d85322262c4396b13ecfff3adb7914~mv2.png/v1/fill/w_215,h_151,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_08d85322262c4396b13ecfff3adb7914~mv2.png)
Extending these methods to long-term activity monitoring in the elderly using smartphones, opened doors to more feasible and convenient approaches to continuously track physical activities.
![image.png](https://static.wixstatic.com/media/3e8f5d_072fe58674344f4da216af9464ab617d~mv2.png/v1/fill/w_241,h_214,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_072fe58674344f4da216af9464ab617d~mv2.png)
My hierarchical approaches were also used for facial expression recognition to addresse the complexities of human emotion detection, demonstrating their versatility and impact.
![image.png](https://static.wixstatic.com/media/3e8f5d_eeac0278b3ff473286aee6d9c1481edc~mv2.png/v1/fill/w_241,h_111,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_eeac0278b3ff473286aee6d9c1481edc~mv2.png)
My work on post-training iterative hierarchical data augmentation has significantly improved deep networks' performance on object detection task
![image.png](https://static.wixstatic.com/media/3e8f5d_cf2794d2799641669b8446f437d5aae2~mv2.png/v1/fill/w_287,h_84,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_cf2794d2799641669b8446f437d5aae2~mv2.png)
The hierarchical feature distillation method I co-designed for zero-shot anomaly detection underscores my contributions to cybersecurity and fraud detection.
![image.png](https://static.wixstatic.com/media/3e8f5d_82840adbcb6b47dcbe558f68e6b1dc7a~mv2.png/v1/fill/w_279,h_101,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_82840adbcb6b47dcbe558f68e6b1dc7a~mv2.png)
Leading the development of a hierarchical transformer for multilingual machine translation showcases my ability to tackle language diversity challenges.
![image.png](https://static.wixstatic.com/media/3e8f5d_c0e6a416b2d442d6927cff5e9f44c132~mv2.png/v1/fill/w_232,h_151,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_c0e6a416b2d442d6927cff5e9f44c132~mv2.png)
Revolutionising electric vehicles’ computational demands, I’ve helped harness the power of nearby vehicles’ resources and deep reinforcement learning for task offloading. This not only curbs energy use but also optimises processing of heavy-duty tasks in ever-changing vehicular landscapes.
![image.png](https://static.wixstatic.com/media/3e8f5d_64a348a3f0384f8f8adf79b29e798757~mv2.png/v1/fill/w_169,h_165,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_64a348a3f0384f8f8adf79b29e798757~mv2.png)
Additionally, my efforts in fair representation learning using uniformly distributed attributes highlight my commitment to ethical AI,
![image.png](https://static.wixstatic.com/media/3e8f5d_2978f946f57a46899adf0344fff49a56~mv2.png/v1/fill/w_205,h_130,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_2978f946f57a46899adf0344fff49a56~mv2.png)
I supervised the development of a Hyperspectral Image Classification method that significantly improves classification accuracy while reducing the need for large labeled datasets. By initially training the model with only 5% of labeled samples and iteratively selecting the most informative samples based on fuzziness, mutual information, and breaking ties, the method fine-tunes the model efficiently, minimizing computational costs.
![image.png](https://static.wixstatic.com/media/3e8f5d_ab20df4f973f4945894d4fb32d917079~mv2.png/v1/fill/w_431,h_200,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/3e8f5d_ab20df4f973f4945894d4fb32d917079~mv2.png)
I supervised the design of a novel method for improving domain generalization of deep models by employing an adversarial reconstruction loss. This approach forces the model to forget style-specific information while retaining class-informative features, resulting in more robust and generalizable models.
Exampler Industrial Consulting
In 2022
In 2022
In 2021
In 2020
In 2019
In 2017
In 2012
In 2010
My strong background and expertise in deep learning and model generalisation were instrumental in the development of modern attention mechanisms, that integrated sparse attention to improve the relevancy and speed of SERPs in Huwei’s petal search project.
My research on improving the generalisation of machine learning models was employed in to improve the performance of the document ranking models, helping them perform well across varied and unforeseen queries, in Huawei’s petal search project.
My model for hierarchical multilingual machine translation, because of its ability to transfer knowledge from majority languages to monitory languages, was adapted to build a translation service for Tatar language
Sarmaya Financials, a unified platform for financial research and portfolio tracking, leveraged my work on hierarchical augmentation and hierarchical zero-shot anomaly detection to enhance their predictive models, operational efficiency and risk management.
Building on my work on domain adaptation, we built a system for department of urban transport that was trained using game generated videos to detect accidents using road side traffic cameras in Kazan.
My work on bone suppression in medical images was used to develop a system which is still assisting medical staff in several hospitals in Kazan in medical image diagnosis.
Building on my methods from my PhD, VisionScape designed and developed their first wrist worn activity tracker in South Korea which was then acquired and launched by SK Telecom.
My work on physical activity recognition (from my Phd) was acquired by Samsung and became an inspiration for the first version of their Samsung Health app back in 2010.
Work Experience
Academic
2022 - Present
Professor, School of Computer Science, University of Hull, United Kingdom
2021-2023
Lead Instructor DAML, Schaffhausen Institute of Technology, Switzerland
2020 - 2022
Professor, Department of Computer Science, Institute of Data Science & Artificial Intelligence, Innopolis University, Russia
2015 - 2019
Associate Professor, Department of Computer Science, Institute of Data Science & Artificial Intelligence, Innopolis University, Russia
2015 - 2015
Academic Consultant, ITU Copenhagen, Denmark
2011 - 2015
Assistant Professor, Department of Computer Science, Ajou University, South Korea
-
Lead the Department’s Machine Learning research group
-
Develop the necessary research infrastructure for state of the art research in Artificial Intelligence
-
Obtaining research funding and provide high quality outputs and publications in Artificial Intelligence
-
Forge partnerships with tech companies and startups to provide students with internship opportunities, collaborative research projects, and potential employment.
-
Collaborate with other universities and research institutions on joint research projects, conferences, and workshops
-
Integrate ethics and responsible AI practices into the curriculum to ensure students understand the societal implications of their work.
-
Supporting and mentoring postgraduate research students in the development of their research careers. Supporting and deliver undergraduate and taught postgraduate curricula, and provide an exceptional experience for students within the department.
-
Play an active role in the life of the Department, by supporting activities around research, teaching, enterprise, student experience and widening participation.
-
Design, develop and teach the module of Dynamic Analysis and Deep Learning
-
Supervise the postgraduate students in their research
-
Develop interdisciplinary programs that integrate data science and AI with other fields such as finance, healthcare, and transport.
-
Create online courses and massive open online courses (MOOCs) to reach a global audience and enhance the university’s reputation.
-
Forge partnerships with tech companies and startups to provide students with internship opportunities, collaborative research projects, and potential employment.
-
Collaborate with other universities and research institutions on joint research projects, conferences, and workshops.
-
Take on leadership roles within the university, such as heading departments or leading research initiatives, to drive the university’s vision forward.
-
Integrate ethics and responsible AI practices into the curriculum to ensure students understand the societal implications of their work.
-
Create and teach undergraduate and graduate-level courses in machine learning, data science, and AI. Ensure that the curriculum is up-to-date with the latest advancements and industry needs.
-
Set up dedicated research labs focused on cutting-edge AI and machine learning research.
-
Apply for grants and secure funding from government agencies, private sector, and international organisations to support research projects.
-
Regularly publish research in high-impact journals and present findings at international conferences to establish the university as a leader in AI research.
-
Mentor undergraduate, graduate, and PhD students, guiding them through their research projects and helping them develop their careers.
-
Provide students with opportunities to participate in cutting-edge research projects, encouraging them to publish papers and present at conferences.
-
Design, develop and teach the module of Pervasive Computing
-
Supervise the postgraduate students in their research
-
Create and teach undergraduate and graduate-level courses.
-
Apply for grants and secure funding from government agencies, private sector, and international organisations to support research projects.
-
Regularly publish research in high-impact journals and present findings at international conferences to establish the university as a leader in AI research.
-
Mentor PhD students, guiding them through their research projects and helping them develop their careers.
Leadership
2023 - Present
Director of PostGraduate Research, University of Hull, United Kingdom
2023 - Present
Director of Hull Machine Learning Research Group, University of Hull, United Kingdom
2016 - 2022
Associate Dean of Academic Affairs, Innopolis University, Russia
2018 - 2022
Director, Institute of AI and Data Science, Innopolis University, Russia
2016 - 2022
Director, Machine Learning Lab, Innopolis University, Russia
2012 - 2015
Co-Lead, Knowledge Intensive Software Engineering Lab, Ajou University, South Korea
-
Develop comprehensive orientation programs to acclimate new students to research processes and resources.
-
Provide ongoing professional development workshops.
-
Pair students with experienced faculty mentors who can guide their research and career development.
-
Encourage peer mentoring and support networks among students.
-
Offer career counseling services to help students navigate academic and non-academic career paths.
-
Organise networking events, and collaborations with industry partners.
-
Create spaces and opportunities for students and faculty to collaborate, share ideas, and work on joint projects.
-
Organise regular research seminars, colloquia, and conferences.
-
Regularly gather feedback from students and faculty to understand their needs and challenges, and use this feedback to continuously improve research support services and policies.
-
Develop and articulate a clear vision and strategic plan for the machine learning research group, aligning with the university's overall mission and goals.
-
Identify emerging trends and opportunities in machine learning to ensure the group stays at the forefront of the field.
-
Foster interdisciplinary collaborations within the university and with external institutions, industry partners, and other research groups.
-
Lead and participate in high-quality research projects that contribute to the advancement of machine learning.
-
Secure funding through grants and partnerships to support research activities, ensuring sustainable growth and development.
-
Attract top talent, including faculty, researchers, and students, to build a diverse and capable research team.
-
Provide guidance and mentorship to junior faculty and students, fostering an environment of continuous learning and development.
-
Efficiently manage the group's resources, including budget, facilities, and equipment, ensuring they are used effectively to support research activities.
-
Develop and implement policies and procedures to ensure the smooth operation of the research group.
-
Regularly assess the group's progress towards its goals and report on achievements and challenges to university leadership.
-
Curriculum Development and Enhancement
-
Faculty Recruitment and Development
-
Student-Centered Learning Environment
-
Industry Collaboration for internship opportunities
-
Ensure that the university has state-of-the-art labs, classrooms, and resources
-
Quality Assurance and Assessment
-
Strategic Planning and Leadership
-
Establish the Institute's Vision and Mission
-
Formulate a long-term strategic plan for the institute
-
Foster a collaborative research environment
-
Create specialised research centers and laboratories focusing on emerging areas in AI and data science
-
Recruit leading researchers, faculty, and PhD students with expertise in AI and data science.
-
Ensure efficient allocation and utilisation of resources
-
Define KPIs to measure the success of research, teaching, and community engagement activities.
-
Regularly review and report on these KPIs
-
Foster a culture of continuous improvement by encouraging feedback, reflection, and adaptation
-
Develop and articulate a clear vision and strategic plan for the lab.
-
Identify emerging trends and opportunities in machine learning.
-
Foster interdisciplinary collaborations within the university and with external institutions, industry partners, and other research groups.
-
Lead and participate in high-quality research projects that contribute to the advancement of machine learning.
-
Secure funding through grants and partnerships to support research activities, ensuring sustainable growth and development.
-
Attract top talent, including faculty, researchers, and students, to build a diverse and capable research team.
-
Provide guidance and mentorship to junior faculty and students, fostering an environment of continuous learning and development.
-
Efficiently manage the group's resources, including budget, facilities, and equipment, ensuring they are used effectively to support research activities.
-
Develop and implement policies and procedures to ensure the smooth operation of the research group.
-
Regularly assess the group's progress towards its goals and report on achievements and challenges to university leadership.
-
Work with the team to identify key research areas in machine learning and software engineering.
-
Develop a long-term research strategy that aligns with the lab's mission and goals.
-
Lead and manage research projects, ensuring they are completed on time and within budget.
-
Supervise and mentor junior researchers, PhD students, and interns.