Researchers at the Kastler Brossel Laboratory (LKB) in Paris have constructed a quantum computer capable of learning complex data without relying on traditional, extensive training methods; instead, the system adjusts only its output layer, a simplification in machine learning protocols. This quantum photonic reservoir computer, equipped with a novel “fading memory” achieved through feedback mechanisms, processes information using entangled light and operates at room temperature. The team, funded by the ERC COQCOoN project and the PEPR OQuLus programme, is exploring applications for forecasting notoriously difficult-to-predict systems. “Reservoir computing” harnesses the natural dynamics of a complex physical system, without training the entire system, enabling machine learning that utilizes the resources of quantum physics. Entangled Quantum Light Enables Reservoir Computing This system, detailed in a recent publication in Nature Photonics, relies on a “reservoir” of entangled light to process information, adjusting only the output layer for learning, a simplification of the typical training process. The experimental protocol was developed by Valentina Parigi’s research team at LKB as part of the ERC COQCOoN project (Continuous Variable Quantum Complex Networks), with support from the PEPR OQuLus programme (Light-based quantum computers in discrete and continuous variables). Unlike algorithms that require exhaustive training, this approach focuses on adapting only the final output stage, reducing computational demands. In this instance, the reservoir is comprised of a multimode quantum state of light, where multiple frequency bands are interconnected through entanglement; researchers utilize a light beam interacting within a non-linear material to create a system capable of information processing. Beyond the fundamental advancement in quantum machine learning, the team is already exploring practical applications for this technology, specifically in forecasting complex time series data. More advanced versions of reservoir computing are being considered for climate and financial market forecasting, highlighting the potential to tackle real-world challenges that demand accurate