Shulou(Shulou.com)06/01 Report--

This article shows you what the machine learning model TensorFlow Quantum based on quantum architecture is like. It is concise and easy to understand. It will definitely brighten your eyes. I hope you can get something through the detailed introduction of this article.

The intersection of quantum computing and artificial intelligence (AI) is expected to become one of the most eye-catching work in the history of science and technology. The emergence of quantum computing may force us to reimagine almost all existing computing paradigms, and artificial intelligence is no exception. However, the computing power of quantum computers has the potential to accelerate many areas of artificial intelligence, which are still impractical. The first step for artificial intelligence and quantum computing to work together is to reimagine the machine learning model as a quantum architecture.

The core idea of TensorFlow Quantum is to interweave quantum algorithms and machine learning programs into the TensorFlow programming model. Google calls this approach quantum machine learning and can do it by taking advantage of some of the latest quantum computing frameworks, such as Google Cirq.

Quantum machine learning

When it comes to quantum computing and artificial intelligence, the first question we need to answer is how the latter benefits from the emergence of quantum architecture. Quantum machine learning (QML) is a broad term that refers to machine learning models that can take advantage of quantum properties. The first QML application focused on refactoring traditional machine learning models so that they could perform fast linear algebra on a state space that grew exponentially with the number of qubits. However, due to the improvement of the computing power of quantum hardware, the development of quantum hardware broadens the field of vision of QML and gradually develops into a heuristic method that can be studied by experience. This process is similar to how the creation of GPU enables machine learning to develop into a deep learning paradigm.

In the context of TensorFlow Quantum, QML can be defined as two main components:

A) Quantum data set

B) mixed quantum model

Quantum data set

Quantum data is any data source that appears in natural or artificial quantum systems. This can be classical data from quantum mechanics experiments, or data directly generated by quantum devices and then input into the algorithm as input. There is some evidence that the application of quantum-classical hybrid machine learning to "quantum data" can provide greater quantum advantages over pure classical machine learning for the following reasons. Quantum data exhibit superposition and entanglement, resulting in joint probability distribution, which may require an exponential number of classical computing resources to represent or store.

Hybrid quantum model

Just as machine learning can generalize models from training data sets, QML can also generalize quantum models from quantum data sets. However, because the quantum processor is still quite small and noisy, the quantum model can not only use the quantum processor to generalize the quantum data. The hybrid quantum model proposes a scheme in which the quantum computer is the most useful hardware accelerator, symbiotic with the traditional computer. This model is well suited for TensorFlow because it already supports heterogeneous computing across cpu, gpu, and TPUs.

Cirq framework

The first step in building a hybrid quantum model is to be able to make use of quantum operations. To do this, TensorFlow Quantum relies on Cirq, an open source framework for invoking quantum circuits on recent devices. Cirq contains basic structures, such as qubits, gates, circuits, and measurement operators, which are necessary for specified quantum computing. The idea behind Cirq is to provide a simple programming model that abstracts out the basic building blocks of quantum applications. The current version includes the following main building blocks:

Circuit (Circuits): in Cirq, Cirquit represents the most basic form of quantum circuit. An Cirq circuit is represented as a set of moments containing operations that can be performed on qubits during some abstract time slip.

Scheduling and equipment (Schedules, Devices): scheduling is another form of quantum circuit that contains more detailed information about the time and duration of the gate. Conceptually, a schedule consists of a set of scheduling operations and device descriptions for running scheduling.

Gate (Gates): in Cirq, the gate abstracts the set of qubits.

Simulators: Cirq contains a Python simulator that can be used for running circuits and scheduling. The simulator architecture can be extended across multiple threads and cpu, which allows it to run fairly complex circuits.

TensorFlow Quantum

TensorFlow Quantum (TFQ) is a framework for building QML applications. TFQ allows machine learning researchers to construct quantum data sets, quantum models, and classical control parameters as tensors in a single computational graph.

From an architectural perspective, TFQ provides a model for abstracting interactions with TensorFlow, Cirq, and computing hardware. At the top is the data to be processed. Classical data is automatically processed by TensorFlow; TFQ adds the ability to process quantum data, including quantum circuits and quantum operators. The next layer is Keras API in TensorFlow. Because one of the core principles of TFQ is native integration with the core TensorFlow, especially with the Keras model and optimizer, this level spans the entire width of the stack. Below the abstraction of the Keras model are our quantum layers and differentiators, which can realize the automatic classification of hybrid quantum classics when connected to the classical TensorFlow layer. Under the layer and delimiter, TFQ relies on the TensorFlow operation, which instantiates the data flow graph.

From an implementation perspective, TFQ follows these steps to train and build the QML model.

Prepare a quantum data set: quantum data is loaded as a tensor and specified as a quantum circuit written in Cirq. The tensor is executed by TensorFlow on a quantum computer to generate quantum data sets.

Evaluate the quantum neural network model: in this step, the researchers can use Cirq to create a prototype of the quantum neural network, which they later embed in the TensorFlow diagram.

Sample or average: this step uses the method to calculate the average of multiple operations involving steps (1) and (2).

Evaluate a classical neural network model: this step uses the classical depth neural network to extract the correlation between the measurements extracted in the previous step.

Evaluate the cost function: similar to traditional machine learning models, TFQ uses this step to evaluate the cost function. This may be based on the accuracy of the classification task performed by the model if the quantum data is tagged, or based on other criteria if the task is not supervised.

Evaluation gradient and update parameters-after evaluating the cost function, the free parameters in the pipeline should be updated in the expected cost reduction direction.

The combination of TensorFlow and Cirq gives TFQ a range of functions, including a simpler and familiar programming model, and the ability to train and execute multiple quantum circuits at the same time.

Efforts to connect quantum computing and machine learning are still at a very early stage. Of course, TFQ represents one of the most important milestones in this field, taking advantage of some of the best IP in quantum and machine learning.

The above is what the machine learning model TensorFlow Quantum is based on the quantum architecture. Have you learned the knowledge or skills? If you want to learn more skills or enrich your knowledge reserve, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.