A new quantum protocol for oblivious transfer (QOT) has been proposed, which could be used in commercially available quantum infrastructures. The protocol involves a sender (Alice) preparing and transferring two one-bit messages to a receiver (Bob), who can choose to learn either one of the messages, but learns nothing about the remaining one. A third party, Trent, serves as a quantum state generator. The protocol is unconditionally secure, as neither Alice nor Bob can interfere with Trent’s state generations. The protocol can tolerate high error rates in quantum computing and quantum communication. The Quantum-aided article was published in Nature.
Quantum Oblivious Transfer Protocol
The article discusses a quantum oblivious transfer (QOT) protocol, which is a cryptographic primitive that is applicable to commercially available quantum infrastructures. This protocol provides a theoretical security guarantee and is designed to work with limited fidelity and quantum capacity.
The QOT protocol involves a sender (Alice) preparing and transferring two one-bit messages to a receiver (Bob). Bob can choose to learn either one of the two messages, but learns nothing about the remaining one. The protocol assumes a three-party model where any party can be dishonest, but the third party cannot collude with the communicating parties. This third party, referred to as Trent, serves as a quantum state generator not directly involved in the computation.
The entire process can be divided into three stages: state preparation, validation and transfer. In the state preparation stage, Trent prepares a sequence of entangled quantum states and sends the entangled pairs to Alice and Bob. The validation stage involves Alice and Bob receiving the corresponding states and randomly choosing some of the states for validation. The final stage is transfer, where Bob measures the quantum states and saves the indices of states whose first bit is chosen.
Quantum-Aided Secure Deep Neural Network Inference
The article also discusses the architecture of quantum-aided secure deep neural network (DNN) inference. The basic component of this architecture is the QOT protocol. The basic operator is securely evaluated with QOT, and the affine transformation is composed with the basic operators. The neural network is implemented with affine transformations as the basic blocks.
The general architecture of quantum-aided DNNs is divided into the operator and network layers. The operator layer consists of QOT and quantum secure communication, which make up the basic operator set, including secure inner product and secure addition. By composing the basic operators, an operator set consisting of affine transformations is created. The network layer of quantum-aided DNNs is comprised of affine transformation blocks. Some layers can remain to be evaluated with classical computing for speeding up, and the rest are evaluated with the quantum protocol to ensure security.
“We claim that QOT is unconditionally secure, because neither Alice nor Bob can interfere with Trent’s state generations, making attacks from Alice or Bob impossible. Meanwhile, as the measurement results of Bob’s or Alice’s alone contain no secret information, Trent gets no information by attacking Alice or Bob. The formal security proof is demonstrated in Methods.”
Authors of the Article
To compose a DNN model with the QOT primitive, the basic blocks of DNN models have to be implemented first. The basic DNN blocks are affine transformations naturally composed of vector inner product and vector addition. The process contains encoding, secure computation, and decoding. All unidirectional communications are assumed to be strictly confidential.
Simulation and Experiment Results
The QOT gate was implemented on real quantum computers on the Cloud and noisy classical quantum simulators. The error rate for a single QOT operation was found to be 0.179. An application-specific quantum computer like a photon computer can further reduce the error rate.
The article also discusses the results of simulations conducted with the Qiskit linear-algebra-based simulator to demonstrate the protocol’s applicability to larger DNNs with special-purpose quantum infrastructures. The noise was imported so that the final oblivious transfer error rate was 5×10^-3, under which the input width of the neural network can be up to 100.
Real-World Application of Quantum-Aided DNNs
The article also discusses the application of quantum-aided DNNs to real-world sensitive data. Simulations were conducted on the common dataset for medical image classification, MedNIST2. A modified AlexNet with two convolutional layers and four fully connected layers was adopted as the classifier. The last two layers were implemented with the quantum-aided protocol. The model was trained for 10 epochs. The classification results and accuracy curve demonstrated that the quantum-aided model has comparable performance with the classical DNN model on real medical images.
The MLC no-go theorem implies that the ideal one-sided two-party oblivious transfer is impossible to be unconditionally secure, with either classical or quantum methods14,16,33. Hence, we adopt a three-party model where any party can be dishonest, but the third party cannot collude with the communicating parties. We will elucidate why this assumption does not violate the requirements of unconditional security. To achieve the concept of oblivious transfer, we refer to this third party as Trent.
“Note that, since neither our protocol itself nor its security proof depends on the low error rate assumption, QOT can tolerate high error rates (noise levels) in quantum computing and quantum communication. Particularly, our protocol passes the error in quantum computing and quantum channels to the next step for DNNs to deal with. Therefore, the overall noise tolerance level only depends on the noise tolerance of DNNs, which can be set by manually introducing noises during DNN training. This is explicitly discussed in Methods.”
Authors of the Article
Quick Summary
A quantum oblivious transfer (QOT) protocol has been proposed that could be used in commercially available quantum infrastructures, offering a theoretical security guarantee. The protocol, which involves a third party known as Trent, could be used to securely implement deep neural networks, with simulations showing that the loss brought by the quantum protocol is insignificant in tasks such as image classification.
- A new quantum oblivious transfer (QOT) protocol has been proposed, which is applicable to commercially available quantum infrastructures.
- The protocol involves three parties: a sender (Alice), a receiver (Bob), and a third party (Trent) who generates quantum states but is not directly involved in the computation.
- The protocol is divided into three stages: state preparation, validation, and transfer.
- In the state preparation stage, Trent prepares a sequence of entangled quantum states and sends them to Alice and Bob.
- In the validation stage, Alice and Bob randomly choose some of the states for validation.
- In the transfer stage, Bob measures the quantum states and sends the indices of states to Alice. Alice then measures the states at the indices and sends an index to Bob.
- The protocol is claimed to be unconditionally secure, as neither Alice nor Bob can interfere with Trent’s state generations.
- The protocol can tolerate high error rates in quantum computing and quantum communication.
- The protocol has been implemented on both real quantum computers and noisy classical quantum simulators.
- The protocol has been used to implement deep neural networks with quantum-aided blocks.
- The protocol has been tested on the binary MNIST classification task, the CIFAR-10 dataset, and the MedNIST dataset for medical image classification. The results show that the loss brought by the quantum protocol is insignificant.
