Quantum algorithm cuts memory needed for future event predictions

Researchers at Nanyang Technological University and the University of Manchester have developed a quantum algorithm that significantly reduces the memory needed to predict future events, addressing a key limitation in applying quantum machine learning. The work offers quantum-algorithmic means to reduce systematic bias while maintaining a quadratic speed-up in calculations of conditional future probabilities, important for estimating the likelihood of events ranging from earthquakes to financial crises. Conventional methods for estimating these probabilities require exponentially increasing memory as complexity grows, a constraint this new approach circumvents. The results advance the timeline for practical quantum technology deployment and illustrate benefits when simulating increasingly complex processes. A quantum model can infer future likelihoods of complex stochastic processes with significantly reduced bias, even when constrained by limited memory, according to work detailed in a recent paper. This achievement addresses a critical hurdle in applying quantum machine learning to real-world prediction tasks, where the amount of memory required to accurately model complex systems often grows exponentially. The core of the innovation lies in designing quantum models that are more expressive than their classical counterparts within the same memory…