|Part 1: Quantum Expectation||What can we reasonably expect from quantum resources?|
|Part 3: Quantum Programing||What does a quantum algorithim look like?|
Part 4: Where can we apply quantum to ML?
|A.||x> =||b> ->|
HHL can do this in exponentially less time than classical
But what do you do with x?
If you try to read out x, you will do log worse than classical
|So either we have to make||x> very small, or only need to know some restricted info about||x>|
N.B. only works for A that are easily simulateable
Can train faster restricted boltzman machines faster on a quantum computer, but encoding the problem you blow away
N.B. All of these make great subroutines in larger algorithims, but we need structure for problems that does not blow away advantages of translating between classical and quantum.
• Part 1: quantum resources, what are they? ○ Qubit, what it is <Whiteboard> ○ Superposition <Whiteboard> ○ Entanglement <Whiteboard> • Part 2: Computing? ○ How does this make a quantum computer ○ What can quantum computers do § Deutch Jorza § Grovers § Shor § Others ○ How are they beating classical? § Something of an open question § Better at noisy functions § Improvement is heavily state dependent § Not strictly better at deterministic functions ○ Common Misconceptions § Superposition does not mean in two states at once § Quantum teleportation teleports information, not anything physical § All quantum systems are non-deterministic, results are probabilistic § http://www.smbc-comics.com/comic/the-talk-3 § DWAVE is not a quantum computer □ It is not proven to be quantum, this is still debated □ Even if it is quantum it is analogue □ It is not a computer, it is an annealer § Quantum crypto, just like classical crypto is not infallible. Just because the protocols work doesn’t mean the implementation does § No such thing as quantum safe, just because it hasn’t been broken with any of the grand total of three algorithms doesn’t make it secure § There may exist algorithms that quantum computers are not good at solving, this does not mean that classical computers will also be bad at solving them • Part 3: QKD ○ How can quantum resources make things secure? ○ QKD protocols? § SPACE!!!!! ○ Why is this not enough? § SIDECHANNELS. § Poor implementations