Duration 27:01
16+
Play
Video

Quantum Machine Learning With TensorFlow Quantum By Nixon Patel, Founder, Qulabs.ai

Nixon Patel
Founder at Qulabs.Ai
  • Video
  • Table of contents
  • Video
Video
Quantum Machine Learning With TensorFlow Quantum By Nixon Patel, Founder, Qulabs.ai
Available
In cart
Free
Free
Free
Free
Free
Free
Add to favorites
57
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speaker

Nixon Patel
Founder at Qulabs.Ai

Senior Executive with 25+ years of experience creating, scaling business growth and technology transformation at quantum computing, voice and digital solution start-ups, public and private companies. Technology Innovator who leads the development and monetization of products to complex problems in business and society through Quantum Computing, Big Data Analytics, Cloud, AI DeepStack, Machine Learning, Mobility and IoT. High-Integrity Leader with success building and inspiring global business, technology, and client-facing teams; recruiting proven executives; and attracting experienced Board members.

View the profile

About the talk

Quantum machine learning is at the crossroads of two of the most exciting current areas of research: quantum computing and classical machine learning. It explores the interaction between quantum computing and machine Learning, investigating how results and techniques from one field can be used to solve the problems of the other. With an ever-growing amount of data, current machine learning systems are rapidly approaching the limits of classical computational models. In this sense, quantum computational power can offer advantage in such machine learning tasks. The field of quantum machine learning explores how to devise and implement quantum software that could enable machine learning that is faster than that of classical computers. Goggle’s open source framework, TensorFlow Quantum (TFQ) ideally facilitates rapid prototyping of hybrid quantum-classical models for classical or quantum data. This framework offers high-level abstractions for the design and training of both discriminative and generative quantum models under TensorFlow and supports high-performance quantum circuit simulators. In this talk I provide an overview of the software architecture and building blocks through several examples and review the theory of hybrid quantum-classical neural networks.

00:10 Intro

04:20 Quantum Computing

09:08 Mathematical representations of qubits

14:06 Quantum algorithms

19:05 QFT design principles

23:01 Getting started with circuit building

Share

Okay. What can I say? Thanks for giving me the opportunity to speak on this very? Breaking topic Quantum machine learning with tensorflow Cointreau and some of the things I've done in the past and some of the things that currently I'm going with Quantum Quantum machine learning on internet communication so far in my start of collapse. Before I dive deep into Quantum Computing and how that relates to a Quantum machine learning tensorflow ice work like to go back and revisit the history of classical electronic

Computing with UCI on the screen here. If you look at the top stop side of the classical Computing developed from 1906 to 40 years to get to a in yet kind of a basic transistor based computer. And then in 1950 was the first IC based in integrated circuit based computer without an alarm 2016-17 and now that are like billions of transistor sitting on a small chip and GPU CPUs whole bunch of things to contrast that with the Hundred Year. Look optical Computing took to make sure

if you look at the quantum Computing cycle is typically start it from the Computing aspect in 1981 with Richard Fineman and he suggested that if you want to simulate anything that natural and definitely it's very hard to do it to a possible computer invented the classical system to perform the simulation on Netflix on the kickoff and then with Shore and go over and a lot of research came in the next couple of decades to water we are now they are we have 5270 cubits Computing available on

Quantum. And today we have plastic surgery operating at 53 from Google library. 72 cubits and a Provost 4000 cubits by 2023 so that's the Map to the next 20 to 25 minutes or what does all this mean and how do I access these Computing Paradigm to solve solve the problems which classical Computing has reached a point of saturation even see examples of how to machine learning from landmarks like tensorflow Quantum Penny Lane and how that can be put together.

so the next slide which I have is an overview of What is a Quantum machine learning I would say it has come up from the machine learning aspect of classical Computing and Quantum Computing aspects walkthrough together. And then currently we have Frameworks like sewer can tensorflow which kind of gives rise to transfer a Quantum from Google which is what we are focusing on today to do the question. typically, it's a Foxwoods of car to mechanic Quantum information Theory computer science and some of the critical

features like superposition in entanglement are the ones that kind of drives Paradigm and why Quantum Computing is I said it was because of Richard Feynman and then a bunch of people like better off and Show Wrangler all those furious people ahead of us to the next level because of the saturation of the woods law and then you know you have this in Helen Lee parallel processing that you can do with Quantum Computing and also having the ability to do entanglement teleportation. So some of these things are also using secure Communication in

case of your bicycle internet. There's always a possibility of Someone Like You eavesdropping got to your pockets while you've been sent from Alice to Bob but because of no clone Theory, Postulate it's pretty secure in Quantum communication, which is currently being applied to something for Computer Solution to look at this diagram like the popsicle computing. Typically how to do in one that's a bit right? So there are two possibilities what you see on the computer,

which is it going to be a bitch has also a zero and one but because of superposition it is very probably stick and it could take any of the superposition State Farm will having Alpha and betta Has amplitudes which kind of give you infinite infinite positionings and infinite possibilities that you can run parallel like, so that's that's the motivation. If you look at it, you can run you can hardly capture an image 300 cubits. That's more number of possibilities that all atoms in the universe gives you motivation why

it's so desirable to use the battle of the Somme an interference that Cardinal Computing allows to solve really hard problem. Coming to now. We just wanted to just walk you through some basic play Cupid is a Quantum Beach and then you know, it's kind of a difference from big that it also takes on a continuous range of values. It's like in the Box fear you saw out the fuse diagram in case of a fascicle. It can only have a position which is Lockport and South Pole

vs. Quantum. It can be any Vector on that fear. So that's the kind of opportunity here. And some of these two properties I talked about the superposition States and makes things probabilistic and continuous. So you have these temperatures Alphas Kirby the square we should be through property should be one and so exploiting the superposition we can have bad luck alley processing capacity diving to the details. I realized that the space available for computation is 2 ^ 10. Just tell

me Walzem time according to Unity operation. So there is a possibility of defining chemical systems. Can I be a Define simulated industrial classical Computing can be possible solutions can be found for example in case of photosynthesis process. How do you come up with a possible new enzyme in case of catalysis so that whole bunch of places where this can be applied mathematical. There's something called rack. rotations with a bracket Ben 10 bracket bracket, so typically a

face your cat is given by Alpha. I know. I50 in this case Abita 132 States so that that's how it is represented in the normal state representative classical registered registers example, if it's a c q bit. Ly bit and now, you know 000 to 1118 East Penn State vs. Quantum, but I don't have anyone say it was all these eight states that can be evaluated. And we talked about entanglement exclusive property. Multistate superpositions import operations can be done Cynthia's lie. There are

things called Gates just like icicle gates with the operator sandgates binary operators in dates for shady operators, which do certain operations which makes competition possible and then they manipulate these gates to do calculations logic gates and their various ways to do that two rotations two different about the medical lab applying different techniques classical Gates your and not more than all these things you're against familiar with and the equivalent Quantum gates are these you have something called.

I did resided identity with just keep stuff you would in the same state X8, which is like a hit that rotates Hubert along the x-axis y-axis z-axis. There were a bunch of other Gates. Of time. I don't think we can go in flight for doing the computations and eventually, you know, bring the logic in the algorithms. I'm just going to that because of lack of time but these are some of the gates signal cubed to cubed Gates. Controlling and feeding entanglement and thereby applying these techniques for drinks calculations. So this is a

Bloch sphere and I told you mention along the z-axis zero and one that's what the classical kind of representation would be when he gets a Quantum. It can be anywhere on the sphere the stage. So there's infinite infinite possibilities the different types of cubits through something called the superconducting qubits spin cubits than there are welcome. There are these two political which is being done by and ions, which is what Microsoft is doing super conducting is Google and IBM

trapped. I know I'm q and Honeyville, so they're different cubits being implemented most important thing to know in this whole aspect Issaquah Highlands the Mount of time the state. Do bettas in the maze in that state for before which you need to finish your computation, right? So that's the important concept a diagram to give you an idea of number of operation before the number of gait speed or different type of like chopped iron vs. Superconducting versus no

city con Diana. Support different type of weird stand in terms of number of gates and number of operations that can be done before Dee Gordon's happens. If you want to be in the top right side, right, so just give you some idea of some of these technologies that are available which would finally go to the ink machine learning in Quantum is to do something cool articles are some functions. Unitree X and Y are two different cubits and you apply some Eritrean

arrive at some results. Show Quantum algorithm how to design them prepare input registers and you know come up with different techniques of these Gates and do a measurement. That's that's that's how the typical I'll go to the world look like this yellow card come Stateside and then that's input State and then you do bunch of operations Olympic you would operation there is the whole because of interference and entanglement you would be able to create some form of

calculations answers in the final one output States would be our final results. And so that's how you arrived at. Adidas algorithms that currently Quantum Quattro speed up you have exponential vs. Polynomial be speeding up and some of the limitations that I try to capture here. So that's just an overview of modern computing. The other aspect is a conceptual framework, which has already been using classical and how does that I love You know come together with the other tools to provide you with a better deal

machine learning self multi-dimensional array lights. Typically you have the concept of it is easier to build a state-of-the-art machine learning models with Charis is a functional Epi support subtraction everywhere Edge devices until 4. So that's the advantage machine bartender tensorflow that will exploit flexible architecture CPU GPU TPU based work on Desktop Service mobile device. So don't come Squanto flu symptoms of low Quantum, which is a combination of

tensorflow and something goes Cirque, which is up. Another tool or another set of circuit simulating to let IBM has come up with so much of a Quantum. It was launched last at the earlier this year and has like a example that I could give us an existing tensorflow estimator API that you can use in the article can essentially abstract many things like go before propagation or backward propagation for programmers and then sticking and a focus on writing the field goal and operations with the dentist if he is attracted to another level of Writing

application simple life so similarly in case of principal Quantum we can use the tensorflow elements answer key elements, which will see further than what we can do with standard answer floor dysfunction like fit and compile like a model. Fit a model Arkham Asylum model with murder compiled or we can end of specified lost functions is it cost functions of lost persons in optimizers. Simple because that's the whole notion of classical date on Quantum data and the quantum data a little bit

of a challenge. Source of possible sources are many you can see Quantum machine learning as a different way to bring interference doing drinking friends are doing as we said entanglement discovering new algorithms particle in a simulation of chemical systems are simulation of anybody's call too many body systems are some of the ways that I get rid of these correlations are computer case of our company two loves of your focusing on Quantum Communication in kontum Internet. Is he building the next gen part of

receivers purifications when you have data on communication Network Netflix on a noisy and you have to distill needs to improve party and some of these are being currently applied. Sephora Contour 2.0 machine learning Design principles for Quantum principal it is important that says in tensorflow in that you need to make sure that you can do before and backward propagation calculations differentiation. Also you need to ensure that That is data that can be generated and should

allow parallel processing. And then execution back in the agnostic meeting you can switch between simulator wanted simulator it because of lack of availability of high-end hardware and is a whole notion of having minimal minimal disruption the new bridge between Zurich and tensorflow is very very soon in the coming slides with initial circuit the album operation, which is a Quantum abilities to create the necessary. So The next thing was to have a quick rundown on the stack, right? So this is a interval Quantum

make a note 40 Beatles live at the North part of the CPU GPU CPU and everything you feel so at the user interface. There are two sets of data. and the principal Quantum supports book They said it's like either circuits are operators. That's what gives you which is equal to Gates Watergate and you can map everything 2 answers and then use the standard tensorflow layers and layers that can be manipulated matrix multiplication and different differentiations to do your Ford Probe backward propagation to created calculus

gradients and through the various operations that can be long or and hardware and Quantum processing unit. And basically that's how the stack is so this is a typical a lame and a very merry toy kind of a scenario of a floworks circuit workflow would be you look at the Y axis u0123 cubic inches in a cubit send a be like kind of operations are applied at Lake ABC. What do you see all different Gates? So that's that's kind of gives you up. Audio of August floor

Lowe's and circle the tool can be used to code the circuits applying these Gates. So that's how this would actually do something like a Quantum algorithm and we just walked around with slides the next slide. Next size of simple circuit you see a simple circuit where cubed to cubed and you have these rotation rotation rotation on the x-axis with the face and lie with Frisbee and is a c not gate which is how you apply like you give up que sera que one of the two

cubits and you get them and then you can call us circuit a method which vs is Wolf connection between these different operations and it's a simple as that is what is the rotational disorder seen our gate logic that has to be done so we can see if we can do a machine-learning on this final space time left. Dark blue as the general and of a circuit in South Ste. 102 cubits apply to model 210 set football team players that stuff to give the data and then let them

and that's a way to use the pipeline for running a typical session to understand how these play work work cited machine learning guide over and environment go to Jackson different box the quantum repair, and then you apply some parameters. Benjamin so you get classical data and then you give it back to that this way you can optimize both are parameters for Quantum circuit and positive circuits. So I think that's that's I think that gives me like few minutes. And so I have some of the applications

that are currently people are working on in terms of chemistry biology into the physics encryption, which is very big communication. That's worth something that we are working on a different aspects of computing. I like that especially a Quantum machine learning aspects. So the radius Frameworks like when I cover tense of a Quantum something by Penny Lane Xanadu from Frameworks are continuously being created to enable people with the background or machine learning and tensorflow kind of exposures are having background python to be able to program

a Quantum machine learning. Using some of these libraries to somebody work we learned today so few minutes. So basic of quantum Computing motivation behind Quantum using quantum computers why hybrid quantum-class tomorrow's work why and how how does influenster work together comes together to give us that chance of a poncho why you sent the report of his work and then we saw how to build a hybrid classical Autumn model for Porter machinery.

Cackle comments for the website

Buy this talk

Access to the talk “Quantum Machine Learning With TensorFlow Quantum By Nixon Patel, Founder, Qulabs.ai”
Available
In cart
Free
Free
Free
Free
Free
Free

Ticket

Get access to all videos “Global Artificial Intelligence Virtual Conference”
Available
In cart
Free
Free
Free
Free
Free
Free
Ticket

Similar talks

Doug Finke
Founder at Quantum Computing Report
+ 3 speakers
Mehdi Bozzo-Rey
Head of Business Development - Quantum at Cambridge Quantum Computing
+ 3 speakers
Shahar Keinan
CEO at Polaris Quantum Biotech
+ 3 speakers
Christopher Bishop
Improvising Careers at Improvising Careers
+ 3 speakers
Available
In cart
Free
Free
Free
Free
Free
Free

Buy this video

Video

Access to the talk “Quantum Machine Learning With TensorFlow Quantum By Nixon Patel, Founder, Qulabs.ai”
Available
In cart
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
561 conferences
22100 speakers
8257 hours of content