According to a Latest Research, Quantum Machine Learning Could Benefit From a Spooky Action That Could Allow exponential Scaling Through Mysterious Quantum Connections

When executed on quantum computers, machine learning, which today enables speech recognition, computer vision, and other tasks, could become even more powerful. The unusual quantum phenomenon is known as entanglement. Einstein labeled “spooky activity at a distance,” which could help remove a significant potential hurdle to applying quantum machine learning.

Quantum computers can outperform conventional ones in various tasks, such as determining a number’s prime factors the mathematical underpinning of modern encryption that protects financial and other sensitive data.

The more qubits in a quantum computer connected by entanglement, the more significant its processing capability can increase exponentially because several particles can influence one other instantly regardless of how far apart they are.

Scientists are still looking at the specific problems where quantum computing might be superior to traditional computing. They’ve recently begun investigating whether quantum computing could aid machine learning, a branch of AI that studies algorithms that better themselves over time.

Simulating quantum systems, such as chemical reactions, is one potential use of quantum machine learning. This could lead to next-generation batteries or new pharmaceuticals. This could comprise building models of the molecules of interest, allowing them to interact, and using real-world experiments as training data to enhance the models.

The so-called “no free lunch” thesis could be a key stumbling barrier for quantum machine learning. When performance is averaged over multiple problems and sets of training data, the theorem states that every machine learning algorithm is as good as, but not better than, any other.

The no-free-lunch theorem states that the average performance of a machine-learning algorithm is dependent on the amount of data it possesses, implying that the amount of data eventually limits machine learning’s performance.

This highlighted the prospect that the amount of training data required by a quantum computer to model a quantum system, for example, might expand exponentially as the modeled system grew larger. This might potentially eliminate quantum computing’s advantage over traditional computing.

Scientists have identified a means to reduce this exponential overhead using a novel quantum version of the no-free-lunch theorem. Their findings, which were confirmed using the Aspen-4 quantum computer from quantum-hardware startup Rigetti, show that increasing entanglement in quantum machine learning can lead to exponential scale-up. The researchers proposed entangling more qubits with the quantum system that a quantum computer is attempting to mimic.

This additional set of “ancilla” qubits can aid the quantum machine-learning circuit is simultaneously interacting with several quantum states in the training data. As a result, even with a small number of ancillas, a quantum machine learning circuit may see a speedup.

Trading entanglement for training states could provide tremendous advantages for training specific quantum systems. Entangling ancilla qubits with the quantum systems utilized in the experiments to provide training data can be tricky. As long as creating entanglement is not exponentially impossible, there can be a benefit in some sense. “Black box uploading,” as the researchers name it, is one possible future application of this technology.

For example, suppose the atom smashers at CERN, the world’s largest particle physics lab, entangled the protons colliding with the detectors used to analyze them and a massive quantum computer (on a billion qubits). In that case, scientists could directly study the Standard Model, which is currently the best explanation for how all known elementary particles behave.

Paper: https://arxiv.org/pdf/2007.04900.pdf

Reference: https://spectrum.ieee.org/quantum-machine-learning

Suggested