From e8db51e977f01061414f05a09d723f3d08f0ec10 Mon Sep 17 00:00:00 2001 From: Fons van der Plas Date: Thu, 3 Sep 2026 16:39:11 +0100 Subject: [PATCH] Fix broken URLs september 2026 --- lectures/Factor Graphs.jl | 2 +- lectures/Intelligent Agents and Active Inference.jl | 6 +++--- tools/DramaBrokenLink.jl | 6 ++++-- 3 files changed, 8 insertions(+), 6 deletions(-) diff --git a/lectures/Factor Graphs.jl b/lectures/Factor Graphs.jl index 32ed8fb6..d2081865 100644 --- a/lectures/Factor Graphs.jl +++ b/lectures/Factor Graphs.jl @@ -709,7 +709,7 @@ keyconcept( md""" ## Final thoughts: Modularity and Abstraction -The great Michael Jordan (no, not [this one](https://youtu.be/cuLprHh_BRg), but [this one](https://people.eecs.berkeley.edu/~jordan/)), wrote: +The great Michael Jordan (no, not [this one](https://youtu.be/cuLprHh_BRg), but [this one](https://en.wikipedia.org/wiki/Michael_I._Jordan)), wrote: > "I basically know of two principles for treating complicated systems in simple ways: the first is the principle of **modularity** and the second is the principle of **abstraction**. I am an apologist for computational probability in machine learning because I believe that probability theory implements these two principles in deep and intriguing ways — namely through factorization and through averaging. Exploiting these two mechanisms as fully as possible seems to me to be the way forward in machine learning." — Michael Jordan, 1997 (quoted in [Fre98](https://mitpress.mit.edu/9780262062022/)). diff --git a/lectures/Intelligent Agents and Active Inference.jl b/lectures/Intelligent Agents and Active Inference.jl index bb127bee..94428ecb 100644 --- a/lectures/Intelligent Agents and Active Inference.jl +++ b/lectures/Intelligent Agents and Active Inference.jl @@ -41,7 +41,7 @@ md""" * Optional - * Noumenal labs (2025), [WTF is the FEP? A short explainer on the free energy principle](https://www.noumenal.ai/post/wtf-is-the-fep-a-short-explainer-on-the-free-energy-principle) + * Noumenal labs (2025), [WTF is the FEP? A short explainer on the free energy principle](https://web.archive.org/web/20260610001030/https://www.noumenal.ai/post/wtf-is-the-fep-a-short-explainer-on-the-free-energy-principle) * A concise, accessible introduction to the Free Energy Principle, aimed at demystifying it for a broader audience—matching the tone and intent suggested by the playful but clear title. * De Vries et al. (2025), [Expected Free Energy-based Planning as Variational Inference](https://arxiv.org/pdf/2504.14898) @@ -151,7 +151,7 @@ Have a look at [Prof. Karl Friston](https://www.wired.com/story/karl-friston-fre ![image Friston presentation at CCN-2016](https://github.com/bmlip/course/blob/main/assets/figures/Friston-2016-presentation.png?raw=true) -In his answer, Friston emphasizes that the first step is to search for food, for instance, a mouse. You cannot eat the mouse unless you know where it is, so the first imperative is to reduce your uncertainty about the location of the mouse. In other words, purposeful behavior begins with [epistemic](https://www.merriam-webster.com/dictionary/epistemic) behavior: searching to resolve uncertainty. +In his answer, Friston emphasizes that the first step is to search for food, for instance, a mouse. You cannot eat the mouse unless you know where it is, so the first imperative is to reduce your uncertainty about the location of the mouse. In other words, purposeful behavior begins with [epistemic](https://en.wiktionary.org/wiki/epistemic#Adjective) behavior: searching to resolve uncertainty. This stands in contrast to more traditional approaches to intelligent behavior, such as [reinforcement learning](https://en.wikipedia.org/wiki/Reinforcement_learning), where the objective is to maximize a value function of future states, e.g., ``V(s)``, where ``s`` might encode how hungry the agent is. However, this paradigm falls short in scenarios where the optimal next action is to gather information, because uncertainty is not an attribute of states themselves, but of *beliefs* over states, which are expressed as probability distributions. @@ -172,7 +172,7 @@ md""" The Free Energy Principle (FEP) is neither a model nor a theory. Rather, it is a principle, that is, a **methodological framework** for describing the information-processing dynamics that *must* unfold in living systems to keep them within viable (i.e., livable) states over extended periods of time. - Think of the processes continuously occurring in our bodies to maintain an internal temperature between approximately ``36^{\circ}\text{C}`` and ``37^{\circ}\text{C}``, regardless of the surrounding ambient temperature. -The literature on the FEP is widely regarded as difficult to access. It was first formally derived as a specific case of the [Least Action Principle](#The-FEP-is-a-Least-Action-Principle-for-"Things") by Friston in his monograph [Friston (2019), A Free Energy Principle for a Particular Physics (2019)](https://arxiv.org/abs/1906.10184), and more recently presented in a more accessible form in [Friston et al. (2023), Path integrals, particular kinds, and strange things](https://doi.org/10.1016/j.plrev.2023.08.016). For a concise and approachable introduction, the explainer by [Noumenal Labs (2025), WTF is the FEP?](https://www.noumenal.ai/post/wtf-is-the-fep-a-short-explainer-on-the-free-energy-principle) is currently the most accessible resource I am aware of. +The literature on the FEP is widely regarded as difficult to access. It was first formally derived as a specific case of the [Least Action Principle](#The-FEP-is-a-Least-Action-Principle-for-"Things") by Friston in his monograph [Friston (2019), A Free Energy Principle for a Particular Physics (2019)](https://arxiv.org/abs/1906.10184), and more recently presented in a more accessible form in [Friston et al. (2023), Path integrals, particular kinds, and strange things](https://doi.org/10.1016/j.plrev.2023.08.016). For a concise and approachable introduction, the explainer by [Noumenal Labs (2025), WTF is the FEP?](https://web.archive.org/web/20260610001030/https://www.noumenal.ai/post/wtf-is-the-fep-a-short-explainer-on-the-free-energy-principle) is currently the most accessible resource I am aware of. In this lecture, we present only a simplified account. According to the FEP, the brain is a generative model for its sensory inputs, such as visual and auditory signals, and **continuously minimizes variational free energy** (VFE) in that model to stay aligned with these observations. Crucially, VFE minimization is the *only* ongoing process, and it underlies perception, learning, attention, emotions, consciousness, intelligent decision-making, etc. diff --git a/tools/DramaBrokenLink.jl b/tools/DramaBrokenLink.jl index a5be51b1..7c6384bf 100644 --- a/tools/DramaBrokenLink.jl +++ b/tools/DramaBrokenLink.jl @@ -18,8 +18,10 @@ const mock_browser_headers = [ const impossible_domains = [ - "research.tue.nl", - "doi.org", + "research.tue.nl", # idk why, but seems reliable anyways + "doi.org", # idk why, but seems reliable anyways + "mitpress.mit.edu", # idk why, but seems reliable anyways + "bmlip.nl", # redirect not understood by HTTP.jl 1.0 ]