#bayesianism

I plan to having a significant focus on #Bayesianism in my advanced #philsci course, next term. How much time should I plan to devote to explaining the basics, assuming that most students lack any background in probability and no math skills?

2025-01-31

I posted this yesterday, but I should have noted that my Cambridge Element ‘Probability and Inductive Logic’ is free to download for the next four weeks. Get amongst it!

doi.org/10.1017/9781009210171

#philosophy #Bayesianism #induction #confirmation #probability #philosophyofscience

Cover image for Antony Eagle, ‘Probability and Inductive Logic’, Cambridge Elements
2025-01-30

My long-in-preparation Cambridge Element ‘Probability and Inductive Logic’ is now available.

doi.org/10.1017/9781009210171

Abstract: Reasoning from inconclusive evidence, or 'induction', is central to science and any applications we make of it. For that reason alone it demands the attention of philosophers of science. This element explores the prospects of using probability theory to provide an inductive logic: a framework for representing evidential support. Constraints on the ideal evaluation of hypotheses suggest that the overall standing of a hypothesis is represented by its probability in light of the total evidence, and incremental support, or confirmation, indicated by the hypothesis having a higher probability conditional on some evidence than it does unconditionally. This proposal is shown to have the capacity to reconstruct many canons of the scientific method and inductive inference. Along the way, significant objections are discussed, such as the challenge of inductive scepticism, and the objection that the probabilistic approach makes evidential support arbitrary.

#probability #induction #bayesianism #philosophy

2023-09-07

bigthink.com/13-8/qbism-quantu

A post promising a series about #quantum #bayesianism. So far so good, this is worth reading to see where it's headed.

Roban Hultman Kramerroban@sigmoid.social
2023-01-06

Beyond the applied ML context, I see a lot of connections to the limitations of #utilitarianism, utilitarian aggregation / #populationEthics, #rationality, #longtermism, subjective #Bayesianism, etc. But would anyone read any of that?

2022-12-19

The No Free lunch theorem implies that we need #Bayesianism

Ok, let me explain. The fact that there cannot be a general purpose algorithm for everything means that we need a prior to solve a certain problem.

Imagine for example that you want to deblur an image, the more a priori you have, the best you will perform. Is it a face or some text? what is the font? what is the language? what is the type or blur (Gaussian kernel, pixelisation, etc)?

#NFL #AI #MachineLearning #Bayes

No Free Lunch Theorem

A diagram showcasing a specialist algorithm being on average the same as a generalist algorithm and how it implies that it has to be worse where it is not specialized.

The inequality for the NFL for a 0-1 loss binary classifier

and below, Bayes' formula

---

A reference to "The Moon Is a Harsh Mistress" by Robert A. Heinlein, with the sentence taken from the text where "There ain't no such thing as a free lunch" appears.
2022-12-17

#introduction

I'm a #mathematician by training and a #datascientist by trade. Living in #munich, #germany. I am a husband to my beautiful wife and father of two amazing children.

I studied math (and some physics) at #LMUMünchen.

I used to work for #mckinsey and #airbnb

I'm interested in #boardgames, #philosophyofscience, #epistemology, #bayesianism, #hierarchicalmodels, #forecasting, #ManyWorlds, #machinelearning, #artificialintelligence, #emergence, #categorytheory

2022-12-16

A question for epistemologists! When Bayesians talk about degrees of belief, do they have any resources to explain why a 0.5 degree of belief ('no idea whether it's true or false') is a worse epistemic position to be in than a 0.99 or 0.01 degree of belief ('very confident about my judgement')? My sense is that the answer is 'no', and that Bayesian epistemology turns inquiry into mechanical rule-following rather than a goal-directed rational process. #Bayesianism #Philosophy #Epistemology

2022-11-27
2022-11-24

🤔 Bayesian Inference (on graphical models) is NP-hard.

But even worst! every epsilon-approximation is also NP-hard.

Which means that the worst case scenario is (almost certainly) exponential.

Good news is, there are some special cases where approximation or exact inference can be performed efficiently.

📘 Check out more in "Probabilistic Graphical Models: Principles and Technique" by Daphne Koller and Nir Friedman

#Bayes #bayesianism #MachineLearning #AI #ML #BayesianInference #Inference

Cover of the book "Probabilistic Graphical Models: Principles and Techniques" by Daphne Koller and Nir FriedmanA figure of the book titled "A reader's guide to the structure and dependencies in this book" showing various subjects covered in the book
2022-11-19

Do you know Judea Pearl?

He bootstrapped the probabilistic approach to AI in his 1988 book
📘 "Probabilistic Reasoning in Intelligent Systems"

He also invented Bayesian Networks and contributed greatly to causality.

Most famous for Pearl's Calculus and the do operator.

🏆 Turing award "for fundamental contributions to artificial intelligence through the development of a calculus for probabilistic and causal reasoning"

#uncertainty #bayes #bayesianism #AI #ML #MachineLearning #researcher

Judea Pearl, born 1936 (aged 86).
Some of his famous work:
Pearl Calculus and Bayesian Networks + some of his books (Probabilistic Reasoning in Intelligent Systems, 1988, Causality: Models, Reasoning, and Inference, 2000 and The Book of Why: The New Science of Cause and Effect, 2018)

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst