PhD Candidate in Explainable AI for Synthetic Media Analysis

Kiran Raja; Gjøvik; On-site; NOK 470,000 - 580,000; Full time

Job summary

NTNU invites applications for a PhD candidate in machine learning at the Department of Computer Science (IDI), NTNU, Gjøvik. The position is a three-year full-time doctoral appointment based at Gjøvik campus.

You will work in a strong international research environment, focusing on authenticating visual content, explainability, and robust evaluation, with opportunities to publish internationally.

Qualifications

• Master's degree or equivalent in computer science, AI, ML, CV or signal processing.
• Admission to the IE faculty's Doctoral Programme is required.

Responsibilities

• Develop machine-learning methods for detecting, localizing, or characterizing synthetic visual content.
• Study representations learned by large pretrained visual or multimodal encoders and adaptation strategies.
• Design explanation methods linking predictions to meaningful evidence.
• Evaluate robustness and transfer to unseen datasets and manipulation processes.
• Investigate uncertainty, calibration, and reliability measures for responsible decisions.
• Publish results in peer-reviewed venues and participate in doctoral training.

Education

Master's degree or equivalent in computer science, artificial intelligence, machine learning, computer vision, or signal processing

Job description

This is NTNU

NTNU is a broad-based university with atechnical-scientific profile and a focusin professional education. The university is located in three cities with headquarters in Trondheim.

At NTNU, 9,000 employees and 43,000 students work to create knowledge for a better world.

You will find more information about working at NTNU and the application process here.

Video: https://youtu.be/Xt-yHCN5QS0

About the position

We have a vacancy for a PhD candidate in machine learning at the Department of Computer Science (IDI), NTNU, Gjøvik. The position is a three-year full-time doctoral appointment and will be based at Gjøvik campus.

Are you motivated to take a step towards a doctorate and open up exciting career opportunities?

As a PhD candidate with us, you will work towards your doctorate in a strong international research environment and gain experience that supports a future career in higher education, research, or knowledge-intensive industry.

The position reports to the Unit Leader of Colorlab.

About the project

The research will address the growing need for reliable methods that can assess the authenticity of visual media and provide understandable evidence for model decisions. The candidate will investigate how general-purpose pretrained visual and multimodal representations can be adapted for analysing authentic, synthetic, and manipulated images or videos.

The work will combine predictive performance with explainability, uncertainty estimation, robustness, and generalization. Attention will be paid to performance across datasets, content sources, generation methods, and real-world transformations such as compression, resizing, and re-encoding.

The final scientific scope will be refined together with the successful candidate and in dialogue with the relevant academic and external research environments.

Duties of the position

• Develop machine-learning methods for detecting, localizing, or characterizing synthetic and manipulated visual content.


• Study representations learned by large pretrained visual or multimodal encoders, including probing, adaptation, fusion, and efficient fine-tuning strategies.


• Design explanation methods that connect predictions to meaningful spatial, temporal, frequency-domain, semantic, or example-based evidence.


• Evaluate robustness and transfer to unseen datasets, content sources, and manipulation processes, including realistic post-processing and distribution shifts.


• Investigate uncertainty, calibration, and reliability measures that can support responsible human decision-making.


• Develop reproducible benchmarks and evaluation protocols covering predictive quality, explanation fidelity and stability, and computational efficiency.


• Publish results in peer-reviewed international venues and present the research at conferences, consortium meetings, and to interdisciplinary audiences.


• Participate in doctoral training, research-group activities, and collaboration with relevant academic and external research environments.

Be prepared for changes to your work duties after employment.

Required selection criteria

• You must meet the requirements for admission to the IE faculty's Doctoral Programme, see Section 6-1 of the PhD regulations for more information.


• You must have a master's degree or equivalent in computer science, artificial intelligence, machine learning, computer vision, or signal processing. The degree must include a substantial independent project equivalent to a master's thesis.


• You must have