19:11 The x1 violates the completeness axiom. It's the square of the amplitude that adds up to 1, not the amplitudes themselves. Should it be something like [sqrt(0.8), sqrt(0.1), 0, sqrt(0.1)]^T or [0.8, 0.1, 0, 0.1]^T/||[0.8, 0.1, 0, 0.1]||^2? 48:00 Does the finite-difference gradient calculation require use to run the system once for each parameter to tune? Would not this be awfully inefficient? Great lecture!
Thanks for a talk. A few questions: - How does the angle encoding using R_z described at 22:00 affect the 0 state? Wouldn't you need a superposition for rotation to actually change anything - 41:47 How would a qubit measurement produce a continuous-valued output between -1 and 1? Doesn't it just collapse to state 0 or 1
Great lecture, thank you. At 9:18 what do you mean when you say there are different possibilities for basis states? I was under the impression that the basis states are the normalized eigenvectors of the operators.
9:18, there are not 'different' possibility for basis states, when measured. Given the operator we're measuring, the basis states are fixed being the normalised eigenvectors of the operators. What the different probability distributions specify, are for different possible OUTCOMES and not BASIS.
at 5:11, I think she interchanged the roles: she meant QC and not CQ as the preferred mode for this lecture series - Quantum computation on Classical data.
Thank you for the clear explanation. I would like to learn in depth about 「higher order Encording」presented in this video, could you tell me references?
Thanks!, this lecture is phenomenal!
You explain very fluently and simply, I feel that everything easily enters my mind. Thank you.
vry nice and simple way to presentaion, thanks for this leacture
19:11 The x1 violates the completeness axiom. It's the square of the amplitude that adds up to 1, not the amplitudes themselves. Should it be something like [sqrt(0.8), sqrt(0.1), 0, sqrt(0.1)]^T or [0.8, 0.1, 0, 0.1]^T/||[0.8, 0.1, 0, 0.1]||^2?
48:00 Does the finite-difference gradient calculation require use to run the system once for each parameter to tune? Would not this be awfully inefficient?
Great lecture!
Thanks for a talk. A few questions:
- How does the angle encoding using R_z described at 22:00 affect the 0 state? Wouldn't you need a superposition for rotation to actually change anything
- 41:47 How would a qubit measurement produce a continuous-valued output between -1 and 1? Doesn't it just collapse to state 0 or 1
Great lecture, thank you. At 9:18 what do you mean when you say there are different possibilities for basis states? I was under the impression that the basis states are the normalized eigenvectors of the operators.
9:18, there are not 'different' possibility for basis states, when measured. Given the operator we're measuring, the basis states are fixed being the normalised eigenvectors of the operators. What the different probability distributions specify, are for different possible OUTCOMES and not BASIS.
woww such a outstanding video ❤
at 5:11, I think she interchanged the roles: she meant QC and not CQ as the preferred mode for this lecture series - Quantum computation on Classical data.
Thank you very much~ I understood it well with an easy explanation~~~
thank you,it will be very helfull to me .but what shall we do in the case of multi classes classification?
Thanks Amira... I love you... It will be very helpfull to my thesis....
Thank you for the clear explanation.
I would like to learn in depth about 「higher order Encording」presented in this video, could you tell me references?
Thank you
Ansatz is a german word. nice to see.