Home

Awesome

Awesome Quantum Machine Learning Awesome

A curated list of awesome quantum machine learning algorithms,study materials,libraries and software (by language).

Main Architecture

Quantum Kernel

In Depth Physics Comparison

Table of Contents

<!-- MarkdownTOC depth=4 --> <!-- /MarkdownTOC -->

<a name="introduction"></a>

INTRODUCTION

<a name="introduction-why-quantum-machine-learning"></a>

Why Quantum Machine Learning?

Machine Learning(ML) is just a term in recent days but the work effort start from 18th century.
What is Machine Learning ? , In Simple word the answer is making the computer or application to learn themselves . So its totally related with computing fields like computer science and IT ? ,The answer is not true . ML is a common platform which is mingled in all the aspects of the life from agriculture to mechanics . Computing is a key component to use ML easily and effectively . To be more clear ,Who is the mother of ML ?, As no option Mathematics is the mother of ML . The world tremendous invention complex numbers given birth to this field . Applying mathematics to the real life problem always gives a solution . From Neural Network to the complex DNA is running under some specific mathematical formulas and theorems.
As computing technology growing faster and faster mathematics entered into this field and makes the solution via computing to the real world . In the computing technology timeline once a certain achievements reached peoples interested to use advanced mathematical ideas such as complex numbers ,eigen etc and its the kick start for the ML field such as Artificial Neural Network ,DNA Computing etc.
Now the main question, why this field is getting boomed now a days ? , From the business perspective , 8-10 Years before during the kick start time for ML ,the big barrier is to merge mathematics into computing field . people knows well in computing has no idea on mathematics and research mathematician has no idea on what is computing . The education as well as the Job Opportunities is like that in that time . Even if a person tried to study both then the business value for making a product be not good.
Then the top product companies like Google ,IBM ,Microsoft decided to form a team with mathematician ,a physician and a computer science person to come up with various ideas in this field . Success of this team made some wonderful products and they started by providing cloud services using this product . Now we are in this stage.
So what's next ? , As mathematics reached the level of time travel concepts but the computing is still running under classical mechanics . the companies understood, the computing field must have a change from classical to quantum, and they started working on the big Quantum computing field, and the market named this field as Quantum Information Science .The kick start is from Google and IBM with the Quantum Computing processor (D-Wave) for making Quantum Neural Network .The field of Quantum Computer Science and Quantum Information Science will do a big change in AI in the next 10 years. Waiting to see that........... .(google, ibm).
References

<a name="basics"></a>

BASICS

<a name="basics-what-quantum-mechanics"></a>

What is Quantum Mechanics?

In a single line study of an electron moved out of the atom then its classical mechanic ,vibrates inside the atom its quantum mechanics

<a name="basics-what-quantum-computing"></a>

What is Quantum Computing?

A way of parallel execution of multiple processess in a same time using qubit ,It reduces the computation time and size of the processor probably in neuro size

<a name="basics-quantum-classical-vs"></a>

Quantum Computing vs Classical Computing

<a name="quantumcomputing"></a>

Quantum Computing

<a name="quantumcomputing-atom-structure"></a>

Atom Structure

one line : Electron Orbiting around the nucleous in an eliptical format

atom

<a name="quantumcomputing-photon-wave"></a>

Photon Wave

one line : Light nornmally called as wave transmitted as photons as similar as atoms in solid particles

Photon wave

<a name="quantumcomputing-elecfluctuation-spin"></a>

Electron Fluctuation or spin

one line : When a laser light collide with solid particles the electrons of the atom will get spin between the orbitary layers of the atom

Spin

<a name="quantumcomputing-states"></a>

States

one line : Put a point on the spinning electron ,if the point is in the top then state 1 and its in bottom state 0

States

<a name="quantumcomputing-superposition"></a>

SuperPosition

two line : During the spin of the electron the point may be in the middle of upper and lower position, So an effective decision needs to take on the point location either 0 or 1 . Better option to analyse it along with other electrons using probability and is called superposition

SuperPosition

<a name="quantumcomputing-superpostion-machinelearning"></a>

SuperPosition specific for machine learning(Quantum Walks)

one line : As due to computational complexity ,quantum computing only consider superposition between limited electrons ,In case to merge more than one set quantum walk be the idea

SuperPosition specific for machine learning

<a name="quantumcomputing-classicalbit"></a>

Classical Bits

one line : If electron moved from one one atom to other ,from ground state to excited state a bit value 1 is used else bit value 0 used

Classical Bits

<a name="quantumcomputing-qubit"></a>

Qubit

one line : The superposition value of states of a set of electrons is Qubit

Qubit

<a name="quantumcomputing-basicgates"></a>

Basic Gates in Quantum Computing

one line : As like NOT, OR and AND , Basic Gates like NOT, Hadamard gate , SWAP, Phase shift etc can be made with quantum gates

Basic Gates in Quantum Computing

<a name="quantumcomputing-diode"></a>

Quantum Diode

one line : Quantum Diodes using a different idea from normal diode, A bunch of laser photons trigger the electron to spin and the quantum magnetic flux will capture the information

Diodes in Quantum Computing1 Diodes in Quantum Computing2 Diodes in Quantum Computing3

<a name="quantumcomputing-transistor"></a>

Quantum Transistors

one line : A transistor default have Source ,drain and gate ,Here source is photon wave ,drain is flux and gate is classical to quantum bits

Quantum Transistors1 Quantum Transistors2

<a name="quantumcomputing-processor"></a>

Quantum Processor

one line : A nano integration circuit performing the quantum gates operation sorrounded by cooling units to reduce the tremendous amount of heat

Quantum Processor1 Quantum Processor2 Quantum Processor3

<a name="quantumcomputing-qram"></a>

Quantum Registery QRAM

one line : Comapring the normal ram ,its ultrafast and very small in size ,the address location can be access using qubits superposition value ,for a very large memory set coherent superposition(address of address) be used

QRAM1 QRAM2

<a name="qcmlbridge"></a>

QUANTUM COMPUTING MACHINE LEARNING BRIDGE

<a name="qcmlbridge-complexNumbers"></a>

Complex Numbers

one line : Normally Waves Interference is in n dimensional structure , to find a polynomial equation n order curves ,better option is complex number

Complex Numbers1 Complex Numbers2 Complex Numbers3

<a name="qcmlbridge-tensors"></a>

Tensors

one line : Vectors have a direction in 2D vector space ,If on a n dimensional vector space ,vectors direction can be specify with the tensor ,The best solution to find the superposition of a n vector electrons spin space is representing vectors as tensors and doing tensor calculus

Tensors1 Tensors2 Tensors3 Tensors4

<a name="qcmlbridge-tensors-network"></a>

Tensors Network

one line : As like connecting multiple vectors ,multple tensors form a network ,solving such a network reduce the complexity of processing qubits

Tensors Network1 Tensors Network2 Tensors Network3

<a name="quantumalgorithmsml"></a>

QUANTUM MACHINE LEARNING ALGORITHMS

<a name="quantumalgorithmsml-qknn"></a>

Quantum K-Nearest Neighbour

info : Here the centroid(euclidean distance) can be detected using the swap gates test between two states of the qubit , As KNN is regerssive loss can be tally using the average

<a name="quantumalgorithmsml-kmeans"></a>

Quantum K-Means

info : Two Approaches possible ,1. FFT and iFFT to make an oracle and calculate the means of superposition 2. Adiobtic Hamiltonian generation and solve the hamiltonian to determine the cluster

<a name="quantumalgorithmsml-qfcm"></a>

Quantum Fuzzy C-Means

info : As similar to kmeans fcm also using the oracle dialect ,but instead of means,here oracle optimization followed by a rotation gate is giving a good result

<a name="quantumalgorithmsml-svm"></a>

Quantum Support Vector Machine

info : A little different from above as here kernel preparation is via classical and the whole training be in oracles and oracle will do the classification, As SVM is linear ,An optimal Error(Optimum of the Least Squares Dual Formulation) Based regression is needed to improve the performance

<a name="quantumalgorithmsml-genetic"></a>

Quantum Genetic Algorithm

info : One of the best algorithm suited for Quantum Field ,Here the chromosomes act as qubit vectors ,the crossover part carrying by an evaluation and the mutation part carrying by the rotation of gates

Flow Chart

<a name="quantumalgorithmsml-hmm"></a>

Quantum Hidden Morkov Models

info : As HMM is already state based ,Here the quantum states acts as normal for the markov chain and the shift between states is using quantum operation based on probability distribution

Flow Chart

<a name="quantumalgorithmsml-bayesian"></a>

Quantum state classification with Bayesian methods

info : Quantum Bayesian Network having the same states concept using quantum states,But here the states classification to make the training data as reusable is based on the density of the states(Interference)

Bayesian Network Sample1
Bayesian Network Sample2
Bayesian Network Sample3

<a name="quantumalgorithmsml-antcolony"></a>

Quantum Ant Colony Optimization

info : A good algorithm to process multi dimensional equations, ACO is best suited for Sales man issue , QACO is best suited for Sales man in three or more dimension, Here the quantum rotation circuit is doing the peromene update and qubits based colony communicating all around the colony in complex space

Ant Colony Optimization 1

<a name="quantumalgorithmsml-caautomata"></a>

Quantum Cellular Automata

info : One of the very complex algorithm with various types specifically used for polynomial equations and to design the optimistic gates for a problem, Here the lattice is formed using the quatum states and time calculation is based on the change of the state between two qubits ,Best suited for nano electronics

Quantum Cellular Automata

<a name="qnn"></a>

QAUNTUM NEURAL NETWORK

QNN 1

one line : Its really one of the hardest topic , To understand easily ,Normal Neural Network is doing parallel procss ,QNN is doing parallel of parallel processess ,In theory combination of various activation functions is possible in QNN ,In Normal NN more than one activation function reduce the performance and increase the complexity

<a name="qnn-perceptron"></a>

Quantum perceptrons

info : Perceptron(layer) is the basic unit in Neural Network ,The quantum version of perceptron must satisfy both linear and non linear problems , Quantum Concepts is combination of linear(calculus of superposition) and nonlinear(State approximation using probability) ,To make a perceptron in quantum world ,Transformation(activation function) of non linearity to certain limit is needed ,which is carrying by phase estimation algorithm

Quantum Perceptron 1
Quantum Perceptron 2
Quantum Perceptron 3 Quantum Perceptron 4 Quantum Perceptron 5

<a name="quantumstatistics"></a>

QAUNTUM STATISTICAL DATA ANALYSIS

quantumstatistics1 quantumstatistics2 quantumstatistics3 quantumstatistics4 quantumstatistics5 quantumstatistics6

one line : An under research concept ,It can be seen in multiple ways, one best way if you want to apply n derivative for a problem in current classical theory its difficult to compute as its serialization problem instead if you do parallelization of differentiation you must estimate via probability the value in all flows ,Quantum Probability Helps to achieve this ,as the loss calculation is very less . the other way comparatively booming is Quantum Bayesianism, its a solution to solve most of the uncertainity problem in statistics to combine time and space in highly advanced physical research

<a name="qpl"></a>

QUANTUM PROGRAMMING LANGUAGES , TOOLs and SOFTWARES

<a name="qpl-all"></a>

All

info : All Programming languages ,softwares and tools in alphabetical order

<a name="quantumhottopics"></a>

QUANTUM HOT TOPICS

<a name="quantumhottopics-deepquantumlearning"></a>

Deep Quantum Learning

why and what is deep learning?
In one line , If you know deep learning you can get a good job :) ,Even a different platform undergraduated and graduated person done a master specialization in deep learning can work in this big sector :), Practically speaking machine learning (vector mathematics) , deep learning (vector space(Graphics) mathematics) and big data are the terms created by big companies to make a trend in the market ,but in science and research there is no word such that , Now a days if you ask a junior person working in this big companies ,what is deep learning ,you will get some reply as "doing linear regression with stochastic gradient for a unsupervised data using Convolutional Neural Network :)" ,They knows the words clearly and knows how to do programming using that on a bunch of "relative data" , If you ask them about the FCM , SVM and HMM etc algorithms ,they will simply say these are olden days algorithms , deep learning replaced all :), But actually they dont know from the birth to the till level and the effectiveness of algorithms and mathematics ,How many mathematical theorems in vector, spaces , tensors etc solved to find this "hiding the complexity technology", They did not played with real non relative data like medical images, astro images , geology images etc , finding a relation and features is really complex and looping over n number of images to do pattern matching is a giant work , Now a days the items mentioned as deep learning (= multiple hidden artifical neural network) is not suitable for that
why quantum deep learning or deep quantum learning?
In the mid of Artificial Neural Network Research people realised at the maximum extreme only certain mathematical operations possible to do with ANN and the aim of this ANN is to achieve parallel execution of many mathematical operations , In artificial Intelligence ,the world intelligence stands for mathematics ,how effective if a probem can be solvable is based on the mathematics logic applying on the problem , more the logic will give more performance(more intelligent), This goal open the gate for quantum artificial neural network, On applying the ideas behind the deep learning to quantum mechanics environment, its possible to apply complex mathematical equations to n number of non relational data to find more features and can improve the performance

<a name="qmlvsdl"></a>

Quantum Machine Learning vs Deep Learning

Its fun to discuss about this , In recent days most of the employees from Product Based Companies Like google,microsoft etc using the word deep learning ,What actually Deep Learning ? and is it a new inventions ? how to learn this ? Is it replacing machine learning ? these question come to the mind of junior research scholars and mid level employees
The one answer to all questions is deep learning = parallel "for" loops ,No more than that ,Its an effective way of executing multiple tasks repeatly and to reduce the computation cost, But it introduce a big cap between mathematics and computerscience , How ?
All classical algorithms based on serial processing ,Its depends on the feedback of the first loop ,On applying a serial classical algorithm in multiple clusters wont give a good result ,but some light weight parallel classical algorithms(Deep learning) doing the job in multiple clusters and its not suitable for complex problems, What is the solution for then?
As in the title Quantum Machine Learning ,The advantage behind is deep learning is doing the batch processing simply on the data ,but quantum machine learning designed to do batch processing as per the algorithm
The product companies realised this one and they started migrating to quantum machine learning and executing the classical algorithms on quantum concept gives better result than deep learning algorithms on classical computer and the target to merge both to give very wonderful result
References

<a name="quantummeetups"></a>

QUANTUM MEETUPS

<a name="quantumdegrees"></a>

QUANTUM BASED DEGREES

Plenty of courses around the world and many Universities Launching it day by day ,Instead of covering only Quantum ML , Covering all Quantum Related topics gives more idea in the order below

Available Courses

Quantum Mechanics for Science and Engineers
Quantum Physics
Quantum Chemistry
Quantum Computing
Quantum Technology
Quantum Information Science
Quantum Electronics
Quantum Field Theory
Quantum Computer Science
Quantum Artificial Intelligence and Machine Learning
Quantum Mathematics

<a name="quantumconsolidatedresearchpapers"></a>

CONSOLIDATED Quantum Research Papers

<a name="quantumconsolidatedresearchpapers"></a>

Recent Quantum Updates forum ,pages and newsletter

License

License

Dedicated Opensources

Dedicated Opensources

Contribution