AI ALIGNMENT FORUM
AF

Infra-Bayesianism

Aug 27, 2020 by Diffractor

Infra-Bayesianism is a new approach to epistemology / decision theory / reinforcement learning theory, which builds on "imprecise probability" to solve the problem of prior misspecification / grain-of-truth / nonrealizability which plagues Bayesianism and Bayesian reinforcement learning. 

Infra-Bayesianism also naturally leads to an implementation of UDT, and (more speculatively at this stage) has applications to multi-agent theory, embedded agency and reflection. This sequence lays down the foundations of the approach.

39Introduction To The Infra-Bayesianism Sequence
Diffractor, Vanessa Kosoy
5y
50
21Basic Inframeasure Theory
Diffractor
5y
14
8Belief Functions And Decision Theory
Diffractor
5y
7
11Less Basic Inframeasure Theory
Diffractor
5y
1
13Inframeasures and Domain Theory
Diffractor
4y
2
17The Many Faces of Infra-Beliefs
Diffractor
4y
5
7Infra-Miscellanea
Diffractor
3y
0
15Infra-Topology
Diffractor
3y
1
17Infra-Exercises, Part 1
Diffractor, Jack Parker, Connall Garrod
3y
1
4Proofs Section 1.1 (Initial results to LF-duality)
Diffractor
5y
0
4Proofs Section 1.2 (Mixtures, Updates, Pushforwards)
Diffractor
5y
0
4Proofs Section 2.1 (Theorem 1, Lemmas)
Diffractor
5y
0
4Proofs Section 2.2 (Isomorphism to Expectations)
Diffractor
5y
0
4Proofs Section 2.3 (Updates, Decision Theory)
Diffractor
5y
0
4[Unlisted]LBIT Proofs 1: Propositions 1-9
Diffractor
5y
0
4[Unlisted]LBIT Proofs 2: Propositions 10-18
Diffractor
5y
0
4[Unlisted]LBIT Proofs 3: Propositions 19-22
Diffractor
5y
0
4[Unlisted]LBIT Proofs 4: Propositions 22-28
Diffractor
5y
0
4[Unlisted]LBIT Proofs 5: Propositions 29-38
Diffractor
5y
0
4[Unlisted]LBIT Proofs 6: Propositions 39-47
Diffractor
5y
0
4[Unlisted]LBIT Proofs 7: Propositions 48-52
Diffractor
5y
0
4[Unlisted]LBIT Proofs 8: Propositions 53-58
Diffractor
5y
0
5Infra-Domain proofs 1
Diffractor
4y
0
6Infra-Domain Proofs 2
Diffractor
4y
0