# US 2021/0342730 A1 · System and method of quantum enhanced accelerated neural network training

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0342730 A1
- Document ID: US20210342730A1
- Status: Published application
- Kind: A1
- Application number: 17/302,402
- Filed: 2021-05-01
- Published: 2021-11-04
- Earliest claimed: 2020-05-01
- Assignees: Equal1.Labs Inc.
- Applicants: equal1.labs Inc.
- Inventors: David J. Redmond; Dirk Robert Walter Leipold
- Web page: https://patentgrep.com/patent/US20210342730A1

## CPC

- G06N 10/60: Quantum algorithms, e.g. based on quantum optimisation, quantum Fourier or Hadamard transforms
- G06N 3/04: Architecture, e.g. interconnection topology
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/0495: Quantised networks; Sparse networks; Compressed networks
- G06N 3/06: Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 3/096: Transfer learning
- G06N 10/40: Physical realisations or architectures of quantum processors or components for manipulating qubits, e.g. qubit coupling or qubit control
- G06N 10/70: Quantum error correction, detection or prevention, e.g. surface codes or magic state distillation
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A novel and useful system and method of quantum enhanced accelerated training of a classic neural network (NN). The quantum system implements an optimizer that accelerates training of the classic NN by exploiting the properties of quantum mechanics and manipulating the quantum system into a state that represents the complete state of the classic NN, including the loss function. The quantum system is then allowed to transition to its “optimum state” and the minimum energy state is read out from detectors and weight updates are calculated and fed back to the classic NN. Mapping and detection helper neural networks learn the characteristics of the quantum system structures. By averaging this over a number of images the learning weight or gradient of descent can be controlled to yield optimum neural network parameters. The time and energy required for training the classic NN as well as for inference is drastically reduced.

## Claims

### Claim 1

A method of quantum enhanced accelerated training of a neural network, said method comprising:
receiving a plurality of activation function tensors corresponding to features across layers in the neural network;
mapping said plurality of activation function tensors of the neural network to energy levels representing a quantum state in a quantum system;
detecting minimum energy states at one or more observation points in said quantum system after said quantum system converges to a minimum total energy; and
determining an update to one or more neural network parameters in accordance with said one or more minimum energy states detected.

### Claim 2 (depends on 1)

The method according to claim 1, wherein said quantum system comprises an array of quantum dots.

### Claim 3 (depends on 1)

The method according to claim 1, further comprising compressing said plurality of activation tensors using an energy based model.

### Claim 4 (depends on 1)

The method according to claim 1, further comprising utilizing a first helper neural network to reduce the number of signals input to said quantum system before quantum operations are performed.

### Claim 5 (depends on 1)

The method according to claim 1, further comprising expanding detected results output from said quantum system into a larger set of said neural network parameter updates utilizing a second helper neural network.

### Claim 6 (depends on 1)

The method according to claim 1, wherein only a certain subset of energy levels are observable in quantum detectors during said detecting.

### Claim 7 (depends on 1)

The method according to claim 1, wherein said detecting comprises achieving quantum read out by performing multiple quantum non-demolition read-outs in sequence to yield nearly complete information.

### Claim 8 (depends on 1)

The method according to claim 1, wherein said one or more neural network parameters comprises weights and/or biases.

### Claim 9 (depends on 1)

The method according to claim 1, wherein detecting comprises performing quantum tomography to measure higher energy state levels by shifting higher energy levels down to said minimum energy state.

### Claim 10

A quantum optimizer apparatus for accelerating training of a neural network, comprising:
a quantum system;
a classic processor coupled to said quantum system and operative to:
receive one or more neural network activation function tensors;
compress said activation function tensor utilizing an energy based model;
map activation energy represented by said energy based model to quantum states in said quantum system;
detect an energy state at one or more observation ports in said quantum system after said quantum system collapses to a minimum total energy; and
determine updates to one or more neural network parameters in accordance with one or more energy states detected.

### Claim 11 (depends on 10)

The apparatus according to claim 10, wherein said quantum system comprises a plurality of quantum dots.

### Claim 12 (depends on 10)

The apparatus according to claim 10, wherein said quantum system comprises a quantum dot array including said plurality of quantum dots organized in a plurality of rows with at least one observation port on either end of each row.

### Claim 13 (depends on 10)

The apparatus according to claim 10, wherein said activation function tensor is compressed utilizing a helper neural network.

### Claim 14 (depends on 10)

The apparatus according to claim 10, wherein a different imposer frequency is assigned to each activation function, and wherein imposer signal durations correspond to activation levels.

### Claim 15 (depends on 14)

The apparatus according to claim 14, wherein a pulse duration of an imposer signal determines a probability ratio between an original energy level and a destination energy level based on a Rabi oscillation process.

### Claim 16 (depends on 14)

The apparatus according to claim 14, wherein said classic processor is further operative to map quantum states back to known imposer values that yield minimum energy values.

### Claim 17

A method of quantum enhanced accelerated training of a neural network, said method comprising:
receiving a plurality of activation function tensors corresponding to features and layers in the neural network;
compressing said plurality of activation function tensors of the neural network utilizing an energy based model to reduce the number of activation function tensors;
mapping said reduced number of activation function tensors signals to energy levels representing quantum state in a quantum dot array incorporating a plurality of quantum dots;
detecting minimum energy states at one or more observation points in said quantum dot array once said quantum dot array converges to a minimum total energy; and
determining an update to one or more neural network parameters in accordance with said one or more minimum energy states detected.

### Claim 18 (depends on 17)

The method according to claim 17, wherein said compressing comprises compressing said plurality of activation function tensors utilizing a helper neural network.

### Claim 19 (depends on 17)

The method according to claim 17, wherein each activation function is assigned an imposer signal frequency applied to one or more quantum dots, and wherein said imposer signal frequency determines an energy level of one or more quantum dots in said quantum system.

### Claim 20 (depends on 19)

The method according to claim 19, wherein a duration of each frequency applied to an imposer is roughly proportional to an activation level.

### Claim 21 (depends on 19)

The method according to claim 19, wherein a pulse duration of an imposer signal determines a probability ratio between an original energy level and a destination energy level based on a Rabi oscillation process.

### Claim 22 (depends on 19)

The method according to claim 19, further comprising mapping quantum states back to known imposer values that yield minimum energy values.

### Claim 23

A quantum optimizer apparatus for accelerating training of a neural network, comprising:
a quantum system;
a first neural network coupled to said quantum system and operative to:
compress a plurality of activation and loss function outputs from a classic neural network utilizing an energy based model to generate a reduced number of activation and loss function outputs;
map the reduced number of activation and loss function outputs to unique quantum state energy levels in said quantum system;
wherein said first neural network is operative to select an optimum choice of frequencies and pulse durations to apply to said quantum system;
a circuit operative to apply said energy level mappings to said quantum system;
a plurality of detectors operative to detect an energy state at one or more observation ports in said quantum system after said quantum system evolves to a minimum total energy; and
a second neural network coupled to said quantum system and operative to generate updates to one or more neural network parameters to said classic neural network in accordance with a plurality of detected energy states thereby training said classic neural network.

### Claim 24 (depends on 23)

The apparatus according to claim 23, wherein said first neural network and/or said second neural network learn one or more characteristics of said quantum system.

### Claim 25 (depends on 23)

The apparatus according to claim 23, wherein said quantum system comprises a plurality of quantum dots.
