PhD Fellow in Multimodal AI for Mapping Forest Biodiversity

1000scholars; Ås; Presencial; NOK 450.000 - 550.000; Jornada completa

Descripción de la vacante

The National Forest Inventory invites applications for a PhD candidate to develop cutting-edge multimodal AI methods for forest biodiversity mapping. The project will integrate remote sensing, forest inventory data, existing maps and climatic data to create unified representations across scales for biodiversity knowledge gaps.

The PhD is a 3-year position based at Ås, with joint supervision from NIBIO and NMBU, enrolling at NMBU.

Formación

• Must pursue a PhD in a relevant field and conduct cutting-edge AI research.
• Experience with interdisciplinary datasets and geospatial data is desirable.
• Strong ability to publish results in scientific venues.

Responsabilidades

• Develop multimodal deep learning methods to integrate heterogeneous forest datasets.
• Investigate learning across different spatial resolutions and modalities.
• Explore foundation-model and self-supervised learning for forest and biodiversity data.
• Publish findings in international journals and conferences.

Conocimientos

Multimodal learning

Representation learning

Self-supervised learning

Foundation models

Educación

PhD candidate in relevant field

Descripción del empleo

Key Information

The National Forest Inventory is looking for a PhD candidate to develop cutting-edge multimodal artificial intelligence methods for forest biodiversity mapping and monitoring. The project will investigate how heterogeneous sources of information-including remote sensing, forest inventory data, existing forest maps, environmental and climatic data, and emerging large-scale AI representations-can be jointly exploited to characterize forests and their biodiversity.

The research will focus on developing multimodal learning approaches that combine complementary forest information across data sources, spatial scales, and time. A central scientific challenge will be to learn integrated representations of forest ecosystems from datasets with very different characteristics, resolutions, coverage, and levels of supervision. These representations will subsequently be used to address important biodiversity-related knowledge gaps, with particular emphasis on mapping old-growth forests and key biodiversity habitats at regional and national scales.

The PhD position is a 3 years position and is located at NIBIO's headquarters in Ås, approximately 30 km south of Oslo.

The candidate will be enrolled in a PhD programme at the Norwegian University of Life Sciences (NMBU) and will become part of the emerging AI for Nature research environment jointly developed by NIBIO and NMBU.

About the position

The candidate will be jointly supervised by advisors from both NIBIO and NMBU.

The position offers a unique opportunity to conduct fundamental AI research while addressing questions of high societal and environmental relevance. The candidate will have access to extensive national forest datasets and geospatial information, work within an interdisciplinary research environment, and contribute to developing the next generation of AI methods for large-scale forest and biodiversity monitoring

To be a current applicant, you cannot have previously held a research fellowship position at NIBIO, and you cannot have a previous doctorate or have previously held a research fellowship position with funding from the Research Council.

If you do not have skills in Norwegian, Swedish or Danish at level A2 upon employment, NIBIO will offer free Norwegian language training with the goal of achieving Norwegian skills at at least this level.

Main responsibilities

The PhD project will investigate new approaches for learning from multiple complementary sources of forest and environmental information. The research will extend beyond approaches based on individual datasets and explore how multimodal artificial intelligence can integrate complementary information from multiple data sources to characterize forest attributes that are difficult or impossible to observe reliably from any single source alone

The main tasks include:

• Developing multimodal deep learning and representation-learning methods for integrating heterogeneous forest and environmental datasets, including remote sensing, forest inventory observations, existing forest maps, climatic and environmental variables, and large-scale AI-derived geospatial representations.
• Investigating approaches for combining information across different spatial resolutions, temporal scales, and data modalities, including situations where some modalities are incomplete or unavailable.
• Exploring foundation-model and self-supervised learning approaches for extracting transferable representations from large-scale forest and Earth-observation datasets.
• Developing AI models for predicting and mapping old-growth forests, key biodiversity habitats, and other ecologically relevant forest characteristics.
• Incorporating ecological knowledge and biodiversity observations into model development and evaluation.
• Evaluating model generalization and transferability across forest types, environmental gradients, and geographical regions.
• Producing regional and nationwide biodiversity-related forest maps and assessing their accuracy and ecological relevance.
• Publishing results in high-quality international scientific journals and conferences and contributing to collaboration between NIBIO, NMBU, and national and international research partners.

Will be evaluated positively

• Experience with multimodal learning, representation learning, self-supervised learning, or foundation models.
• Experience with geospatial, remote sensing, and citizen science datasets.
• Experience working with heterogeneous datasets originating from different sensors or data sources.
• Familiarity with forest ecology, biodiversity or natural-resource applications.
• Experience developing reproducible research software and working with high-performance or GPU computing environments.
• Experience publishing or contributing to scientific articles or conference papers.

Personal qualifications

• Good interpersonal and communication skills
• Strong analytical and problem-solving skills
• Ability to work under pressure and interact with demanding users
• Commitment and enthusiasm for working as part of an ambitious research team
• Ability to collaborate with both internal and external experts with diverse academic backgrounds and skills