Demonstrating Quantum-Assisted Binary Classification Using a Quantum Annealer

2912 words (12 pages) Essay in Engineering

18/05/20 Engineering Reference this

Disclaimer: This work has been submitted by a student. This is not an example of the work produced by our Essay Writing Service. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.

1. Summary

This document is intended to propose the final area of research regarding the project  that is going to be explored, providing an outline of the direction and method that is to be undertaken, the milestones that are to be achieved and the key stakeholders involved throughout this project.

Get Help With Your Essay

If you need assistance with writing your essay, our professional essay writing service is here to help!

Find out more

The purpose of this report is to provide the reader with a complete understanding of the direction of the field being explored, the relevant literature that will support this research and a high-level plan that will be followed in-order to complete that work. This report further sets the goals of this research and defines what are the implicit motivations for carrying out this study. 

A brief description is mentioned providing the reader with a general overview of the background of the field and the reasons to choose this particular field of study. This is followed by some of problems that will be faced during the project, moving on to the objectives and goals in mind while undertaking this research.

This report then precisely breaks down the goals into smaller tasks that are going to be undertaken to formalize the scope of work for this project. With every project there is a preliminary safety assessment that must be undertaken, and the section that follows describes all the risks, hazards and the precautions that will be taken in-order to avoid them.

A Gantt chart is then created to give the reader a clear idea of the different phases of the project and how the major goals will be achieved as project moves along. The key stakeholder and personnel are further mentioned. Lastly, the risks that are associated with the project are evaluated and illustrated.

2. Brief Description

This project aims to use a quantum annealing machine to run a binary classification problem.
 

Improving the effectivity of calculations has been an objective as far back as PCs were begun to use for tackling critical issues. The unpredictability of calculations is grouped into different classes of complexity, for example, quadratic, exponential and so forth. These classes help in evaluating the time or memory required to run the calculations based on information size. Albeit an enormous advancement is made in creating possible answers for issues by and large, there are still some difficult issues that require exponentially more assets as the information size develops. The ‘travelling sales rep issue’ is an outstanding guide to this case. To such issues, any polynomial unpredictability development arrangement would be considered as an effective arrangement, however an optimal solution is to be developed.
 

Traditional PCs work by encoding information to discrete states: 0 and 1. Traditional PCs, for example, PCs, servers, and cell phones are essentially Turing complete – they can reproduce a Turing machine. Interestingly, quantum bits might be caught condition of 0 and 1 now and again. A proportionate Turing complete quantum computer is simply hypothetical at the hour of composing; however, the D-wave Quantum Annealer has proved an evidence of a “Quantum Advantage”.
 

Quantum Annealing is a computational innovation responsible in solving challenging optimisation issues. It seeks the minima of a function in a complex space.  Physically, the cost capacity encodes as the system’s energy. The calculation continues by setting up the framework in a quantum superposition of every conceivable arrangement in the arrangement space, all similarly plausible, consequently starting an exceptionally quantum parallel handling. The framework at that point is developed in time until they looked for negligible vitality setup is overwhelmingly plausible. As a component of this procedure, the framework has the plausibility of quantum burrowing through tall, slender boundaries in the vitality scene to escape neighbourhood minima in under exponential time.
 

The Quantum Annealing (QA) processor chips that D-wave fabricates uses quantum properties to realise the benefits of QA calculations in hardware. To discover ideal arrangements, qubits are placed into their most reduced state, in which each is in a quantum superposition of both ‘on’ and ‘off’. Magnetic fields interact at that point, and delicately push this uncertain state towards a more certain one — process known as quantum annealing. The state progresses while keeping up its low end with the ultimate objective that when it separates, it should leave qubits in the best arrangement for dealing with that particular issue. Since the framework filters each conceivable answer without a moment’s delay, in principle it is a quicker method to determine issues that get exponentially harder with each additional variable.

3. Objectives

The desired goals and objectives are to prove there exists a quantum advantage. That is the underlying objective. However, taking into account the timescale and resources available, a more achievable target is to successfully run a binary classification model on a quantum annealer in hopes to, at a basic level, prove some of the positives in using quantum annealing. This would prove as a stepping stone to solving take to more challenging machine learning optimisation algorithms in various applications across industry later on. As  Alan Turing put it: We can only see a short distance ahead, but we can see plenty there that needs to be done.

The specific and measurable goals are as follows:

• Goal 1:  Identify and solve a problem related to a non-ferromagnetic ising spin glass model.

• Goal 2: To successfully write a script and train a classification model based of a Quantum-Assisted Machine Learning algorithm.

• Goal 3: To successfully identify which binary classification problem would be suitable with the current resources available (16 bits).

  •      Goal 4: To successfully apply the knowledge of the D-wave Quantum Annealer, the QAML algorithm in order to apply it on classification problem and produce a deliverable result.

 

4. Scope of Work

The research will start with a basic understanding of programming the D-wave computer. Additionally, problems of varied difficulty of the Ising model will be solved, and a review of the capabilities of the D-Wave machine will be tested to scope out the challenges in building and training a binary classification model. By the end of this project, a case study in training a basic classifier using a Quantum Annealer would be presented.

The follow section presents some preliminary research undertaken regarding the D-wave stack since the start of this month.

D-Wave Technology Stack
 

A D-Wave platform comprises two main components:

A processor chip (“the hardware”) that solves Ising Model (IM) problem instances by physical realization of a quantum annealing algorithm. Thee chip is mounted in a dilution refrigerator and supercooled to a target operating temperature below 20mK, which is necessary to achieve quantum effects. The chip subsystem is housed within many layers of shielding to protect against various types of environmental noise.

A conventional (Intel) front end server connected to the chip via control lines and an I/O subsystem. The front end receives instructions from the user and is accessed using a cloud computing model that supports job queuing and scheduling. The front end sends a problem instance to the hardware and receives a set of solutions in reply.

Empirical work has so far shown that D-Wave’s approach to quantum computation is viable, which opens up exciting new directions for bringing quantum computation to bear in real-world applications. A great deal of information and insight can be gained from formulating, implementing, and analysing algorithms running on D-Wave platforms.

The D-Wave QPU is a cross section of interconnected qubits. While some qubits associate with others by means of couplers, the D-Wave QPU isn’t completely associated. Rather, the qubits interconnect in an engineering known as Chimera.

 

Figure 1: Chimera graphs are composed of 8-qubit cells featuring bipartite connectivity. Each cell’s partition is connected to another partition in the adjacent cells.

This information naturally precedes the question, “Can any Ising Model problem be mapped to the Quantum Annealer?” The answer is no, there are limitations to what the D-wave QPU can process. This is because of two reasons in particular:

1. If the number of qubits available in the hardware is lesser than the variables.

2. The Chimera graph connections (seen in figure 1) are not a fully connected graph.

5. Preliminary Assessment of Safety Risks
 

Causes of Safety Hazards
 

1. Postural problems

Occurs due to negligence and/or insufficient planned breaks.

2. Visual problems

Caused by negligence, glare due to external light, or an uncorrected eye issue.

3. Fatigue and stress

Augmented by the stress of the project and problems described in 1 & 2.

4. Insufficient space

Which leads to lack of postural changes.

5. Insufficient internal and external light

Use the keyboard backlight if there is a lack of external light. Take measures to avoid straining eyes.

6. Excessive Noise

Caused by the laptop that may distract thus impairing concentration or preventing normal conversation.

7. Insufficient ventilation

Due to the heat emitted by the laptop which leads to unexpected changes in temperature and humidity.

Overcoming Safety Hazards

 

1. Planned “Time-Off”

It is obvious that the more individuals work at a presentation screen, the more exhausted they become. Planned breaks can diminish weakness and improve efficiency.
 

2. Keyboard & Mouse

All tasks should be done without the unnatural action of twisting. Take necessary precaution if the keyboard & mouse is used a lot.
 

3. Screen Interaction

The laptop must be placed at a comfortable distance and directly in front during intense reading.
 

4. Chair

The Chair must be adjusted at the right height for the vitality your wrists, neck, forearms and feet.

 

5. Worksurface

The work surface should be large enough to allow for some flexibility. The surface should not be too shiny to avoid reflection. 
 

6. External Light

Avoid hunching your shoulders at all times. Additionally, ensure there is no bright glare sources in your field of view.


 

6. Bibliography of Relevant Literature

  • [1] Steven H. Adachi, Maxwell P. Henderson, “Application of Quantum Annealing to Training of Deep Neural Networks.” Lockheed Martin Corporation, Palo Alto, California, USA, https://arxiv.org/abs/1510.06356 (2015).
  • [2]Marcello Benedetti, John Realpe-Gomez, and Alejandro Perdomo-Ortiz, “Quantum-assisted helmholtz machines: A quantum-classical deep learning frame- work for industrial datasets in near-term devices,” arXiv:1708.09784 (2017).
  • [3] Aniruddh Goteti, “Text Classification using Quantum Machine Learning: A Review”
    Blekinge Institute of Technology, International Journal of Engineering Science and Computing, Sweden (2018).
  • [4] Rasdi Rere et al, “Simulated Annealing Algorithm for Deep Learning L.M, ” The Third Information Systems International Conference. Procedia Computer Science 72 (2015).
  • [5] Alejandro Perdomo-Ortiz, “Opportunities and challenges for quantum-assisted machine learning in near-term quantum computers, ” Quantum Artificial Intelligence Lab, NASA. (2018)
  • [6] Alex Mott et al, “Solving a Higgs optimization problem with quantum annealing for machine learning” Macmillan Publishers Limited, part of Springer Nature (2017)
  • [7] Helmut G. Katzgraber,“ Seeking Quantum Speedup Through Spin Glasses: The Good, the Bad, and the Ugly,” D-Wave Systems, Inc. (2015)
     

7. Timetable

ACTIVITY

IMPLEMENTATION TIME

 

October

November

December

January

February

March

1.  Comprehensively understanding and testing the D-Wave QPU Architecture: Chimera
 

Start

Complete

1.1.  Design and implement training program for

the Chimera architecture
 

Complete

1.2. Design training modules
 

Start

Complete

1.3. Solve an example of ising spin glass problem
 

Start

Complete

2. Search for Binary Classification Problems feasible for simulator
 

Start

Complete

3.  Comprehensively understanding and testing QAML algorithms
 

Start

Complete

3.1. Writing good test QAML code to comprehensively understand the algorithm
 

Start

Complete

3.2. Solving a smaller QAML task to build an understanding the embedding problem

 

Start

Complete

4. Solving a binary classification problem writing a QAML algorithms simulated on the D-wave Quantum Annealer

Start

Complete

Detailed information on the expected timetable for the project. The project is broken into 3 phases and a schedule is provided for each phase.

Description of Work

Start and End Dates

Phase One

Comprehensively understanding and testing the D-Wave QPU Architecture: Chimera

09/10/19 – 29/11/19

Phase Two

Comprehensively understanding and testing QAML algorithms

30/11/19 – 19/01/19

Phase Three

Solving a binary classification problem writing a QAML algorithms simulated on the D-wave Quantum Annealer

20/01/19 – 30/03/20


A Gantt chart has been created for a more detailed project timetable
 

8. Key Personnel

The key personnel who will be responsible for completion of the project, as well as other personnel involved in supervising the project.

Personnel Role

Personnel Name

Researcher

Mr. Rowan Shah

Project Supervisor

Dr. Paul Warburton

Project Mentor

Mr. Daniel O’ Connor

Project Co-Ordinator

Dr. Arni McKinley

9. Evaluation

In this section, potential risks to the success of this project are described.

1. Acquisition of items critical to project success (e.g., software resources) could delay in the procurement process.

2. Risks with the software (the development platform) chosen to perform project development. e.g. inability of the D-wave Quantum annealer to handle the workload required to complete the project.

3. Project supervisor or mentor are not available. (Account situations such as vacation, training, travel, and meetings).

4. Imperfect project completion resulting from a mandatory completion date for the project.

5. If the software is prone to bugs and is slow, this can affect development, especially during the coding and testing stages.

6. Software available supplied by companies/universities over which the project supervisor does not have direct control.

7. Researcher is unfamiliar with the environment being used (e.g. the technology is unfamiliar)

Cite This Work

To export a reference to this article please select a referencing style below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please:

Related Lectures

Study for free with our range of university lectures!