The 2027 edition is an in-person conference organised by Visual Intelligence and the UiT Machine Learning Group, with peer-reviewed full papers published in the Proceedings of Machine Learning Research, non-archival extended abstracts, an industry event and a Diversity in AI event. Topics span general deep learning (architectures, generative models, deep reinforcement learning), related machine learning (active, online, self-supervised learning and time-series analysis), optimisation, Bayesian and variational methods, graphical models and explainability, applications to vision, language, signals, speech and audio, and deep learning for the sciences, including biology and medicine, environment and ecology, and physics. The main conference runs 12-14 January 2027; the surrounding NLDL Winter School (11-15 January), part of the NORA Research School and worth 2.5 ECTS, offers tutorials on agentic AI and AI-assisted research, deep ordinal representations for trustworthy AI, evaluating large language models beyond accuracy, and Arctic AI-Earth system modelling and evaluation.