Conference Description
Key Takeaways
- Five-day international research training school focused on recent advances in deep learning
- Designed for graduate students, postgraduates, researchers, and industry practitioners working in artificial intelligence and machine learning
- Hybrid format offering both in-person attendance in Orléans, France, and full live online participation
- Programme includes parallel academic courses, industrial sessions, open research presentations, and a machine learning hackathon
- Co-organised by Université d’Orléans, Collège Doctoral Centre-Val de Loire, and IRDTA
Introduction
DeepLearn 2026, the thirteenth edition of the International School on Deep Learning, brings together academics, researchers, and industry professionals for an intensive week of training on the latest developments in deep learning. Running from July 20–24, 2026, in Orléans, France, the event addresses the persistent challenge facing practitioners in this field: keeping pace with a discipline where foundational techniques and state-of-the-art methods evolve rapidly. The programme combines theoretical instruction with practical application, offering participants the opportunity to learn from leading experts while engaging directly with real-world machine learning problems.
About DeepLearn 2026
Now in its thirteenth year, DeepLearn has established itself as a recurring fixture in the deep learning education calendar. The event operates as a research training school rather than a traditional conference, prioritising structured learning and skills development over paper presentations. This format reflects the growing recognition that deep learning expertise requires continuous education, as techniques that represented the cutting edge just two or three years ago may now be superseded by more effective approaches.
The 2026 edition is hosted in Orléans, a city in central France with a well-established university presence. Université d’Orléans serves as a co-organiser alongside Collège Doctoral Centre-Val de Loire and IRDTA, the latter being the primary coordinating body behind the DeepLearn series. This institutional backing provides participants with access to academic infrastructure while maintaining the event’s international character.
Participation is available in two modes: in-person attendance in Orléans or full live online access. While the organisers accommodate remote participants with complete access to sessions, they emphasise the particular value of face-to-face interaction for networking and collaborative learning. This hybrid approach has become increasingly common in technical education, balancing accessibility with the recognised benefits of physical presence.
Programme Structure and Learning Format
The DeepLearn programme is built around parallel courses that run throughout the week, allowing participants to construct individualised learning paths based on their interests and existing knowledge. Rather than following a single prescribed curriculum, attendees move between sessions according to their priorities. This flexibility acknowledges that participants arrive with varying backgrounds and objectives—a PhD student exploring neural architecture search has different needs than an industry engineer seeking to implement production-ready models.
Academic lectures form the core of the programme, delivered by researchers working at the forefront of deep learning theory and methodology. These sessions address both foundational concepts and recent advances, providing context for understanding why certain approaches have gained traction while others have fallen out of favour. The theoretical grounding is complemented by practical demonstrations that illustrate how abstract techniques translate into working implementations.
An industrial session brings perspectives from practitioners applying deep learning in commercial and operational contexts. This component addresses a frequent criticism of academic training: that it can become disconnected from the constraints and requirements of real-world deployment. Industry presentations typically cover topics such as scaling challenges, infrastructure considerations, and the practical trade-offs involved in moving from research prototypes to production systems.
Open sessions provide a forum for participants to present their own research in progress. These slots serve multiple purposes: they give early-career researchers experience presenting to knowledgeable audiences, create opportunities for feedback from experts in the field, and facilitate connections between people working on related problems. For many attendees, these informal exchanges prove as valuable as the formal instruction.
The hackathon component introduces a competitive element, challenging participants to tackle machine learning problems within constrained timeframes. Hackathons have become a standard feature of technical training events because they force participants to apply newly acquired knowledge under pressure, revealing gaps in understanding that passive learning might not expose. They also encourage collaboration, as teams typically form across institutional and national boundaries.
Deep Learning in 2026: Why Continuous Education Matters
The deep learning field presents unusual challenges for professional development. Unlike more stable technical disciplines where foundational training remains relevant for decades, deep learning practitioners must continuously update their knowledge to remain effective. Architectural innovations, training techniques, and best practices shift regularly, driven by ongoing research and the computational resources that enable experimentation at scale.
This rapid evolution creates a knowledge gap that affects both individuals and organisations. Researchers risk pursuing approaches that have been superseded, while industry teams may miss opportunities to improve their systems using newer methods. Events like DeepLearn address this gap by providing concentrated exposure to current thinking, delivered by people actively contributing to the field’s development.
The intersection of academic research and industrial application has become increasingly important as deep learning moves from experimental technology to operational infrastructure. Techniques developed in research laboratories now power systems handling significant commercial and societal functions. This transition demands practitioners who understand both the theoretical foundations and the engineering realities of deployment—precisely the combination that DeepLearn’s mixed academic and industrial programme aims to develop.
Who Should Attend
DeepLearn 2026 is designed to accommodate participants across a range of career stages and professional contexts. The absence of formal prerequisites means the event welcomes those relatively new to deep learning alongside experienced practitioners seeking to update specific areas of their knowledge.
Graduate and postgraduate students form a core constituency, particularly those pursuing research in machine learning, computer vision, natural language processing, or related fields. For doctoral candidates, the event offers exposure to research directions that may inform their thesis work, along with networking opportunities with potential collaborators and future employers.
Industry practitioners—including data scientists, machine learning engineers, and research scientists working in commercial settings—benefit from the academic rigour that distinguishes DeepLearn from vendor-focused training. The industrial sessions and hackathon provide direct relevance to their professional responsibilities, while the academic content offers deeper understanding of the techniques they implement.
Academic researchers and faculty members attend both to update their own knowledge and to identify emerging topics suitable for incorporation into their teaching and research programmes. The international participant base also creates opportunities for establishing collaborative relationships across institutions.
Bridging Research and Practice
One of the persistent tensions in deep learning education is the gap between research advances and practical implementation. Academic papers often present results achieved under idealised conditions, while practitioners face constraints around data quality, computational budgets, and deployment requirements that research settings rarely address. DeepLearn’s combination of academic and industrial perspectives attempts to bridge this divide, giving participants insight into both what is theoretically possible and what is practically achievable.
The hackathon reinforces this practical orientation by requiring participants to produce working solutions rather than theoretical analyses. Time pressure and real datasets expose the difference between understanding a technique conceptually and implementing it effectively—a distinction that matters considerably in professional contexts.
For organisations investing in their teams’ development, events like DeepLearn offer returns that extend beyond the immediate knowledge transfer. Participants return with updated mental models of the field, awareness of emerging techniques worth investigating, and professional connections that can prove valuable for future recruitment or collaboration. In a discipline where talent is scarce and knowledge becomes outdated quickly, such investments in continuous education represent a practical necessity rather than a discretionary benefit.