Member-only story
All about Data Encoding for Quantum Machine Learning
Quantum Data embedding Methods for Quantum Machine Learning
Quantum Computer expects data in Quantum state for processing. NISQ (Noisy Intermediate Scale Quantum) devices contain a limited amount of qubits, and these qubits are stable for a concise period of time. The first step in Quantum Machine Learning is to load classical data by encoding it into the state of the Qubits. This process is also known as Quantum Data encoding or embedding and is an important step in Quantum state preparation. Classical data encoding for Quantum computation plays a critical role in the overall design and performance of the Quantum Machine Learning algorithm (QML).
The representation must be compact and use only a few qubits and few quantum gates to use current NISQ devices. Qubits do not just decay fast, and the quantum gates are error-prone, too, limiting the number of operations to prepare the quantum state, which must be small.
Encoding can be categorized into two categories :
- digital encoding is the representation of data as qubit strings
- analogue encoding represents data in the amplitudes of a state
If data has to be processed by arithmetic computations, a digital encoding is preferable. While for machine learning algorithms, analogue…
