Quantum mechanics, which is the study of the behavior of sub-atomic particles, provides a way to enhance the use of machine learning to resolve inherently complex problems around optimization, product development and real-time decision making.
This quantum-inspired enhancement, known as a tensorized neural network, is helping companies realize savings in training for a wide range of applications, including self-driving vehicle simulations, natural language processing and even generative AI like ChatGPT.
Tensor networks address one of the constraints facing companies investing in machine learning: the scale of resources required to model and train systems representing complex, multi-dimensional entities like fluid dynamics or financial systems. More complexity represented in the modeling demands correspondingly higher resources in terms of energy, memory size and computing time for training with traditional neural networks on classical computers (see "Compute Trends Across Three Eras of Machine Learning").
Quantum-Inspired Tensor Networksne of the key features of a neural network is its robustness. When neural networks are trained on slightly different datasets, robustness means that they perform similarly. Think of it as wanting your self-driving car to know what to do if a red car is parked on the motorway, even if it has only seen blue cars parked on the motorway during its training.
One way to add robustness is to make neural networks highly redundant by adding more nodes (the building blocks of neural networks), connections and layers. Highly complex networks will tend to behave similarly, even if they have been trained slightly differently. Yet this robustness increases the number of variables to tune, leading to a trade-off between a model’s reliability and the added cost in terms of more time and computational memory required for training.
Read full Forbes article by Sam Mugel here.