Gödel Machines: Fully Self-referential Optimal Universal Self-improvers

By J. Urgen Schmidhuber. 2007

149 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computational Theory and Mathematics

Citations in the corpus

Deep learning in neural networks: An overview (2014, methodology): “…Graves, & Schmidhuber, 2007b; Graves& Schmidhuber, 2009) trained by Connectionist Temporal Classification (CTC) (Graves, Fernandez, Gomez, & Schmidhuber, 2006), a gradient-based method for finding RNN weights that maximize the probability of teacher-given label sequences, given (typically…”
PowerPlay: Training an Increasingly General Problem Solver by Continually Searching for the Simplest Still Unsolvable Problem (2011, methodology): “Unlike the task search of theoretically optimal creative agents (Schmidhuber, 2006a, 2010) (Section 7.4), PowerPlay’s task search is greedy, but at least practically feasible.”
Some Considerations on Learning to Explore via Meta-Reinforcement Learning (2018, background): “Meta RL goes by many different names: learning to learn, multi-task learning, lifelong learning, transfer learning, etc [25, 26, 22, 21, 23, 12, 24, 39, 37].”
Evolved Policy Gradients (2018, background): “For example, (41; 40) consider self-modifying learning machines (genetic programs).”
Artificial General Intelligence (2016, methodology): “Omitting these variables from the task model is tantamount to making the untenable assumption that these resources are infinite [12].”
Self-Modification of Policy and Utility Function in Rational Agents (2016, unknown): “Keywords”
Exploration from Generalization Mediated by Multiple Controllers (2013, background): “The prospect of rich links between these formal methods and intrinsic motivational concepts such as curiosity from the same group (e.g., Schmidhuber 2006) is enticing.”
Goal-Directed Planning via Hindsight Experience Replay (2022, result): “While meta-learning has been investigated early in the literature (Schmidhuber, 1987; 2007), recent work has shown impressive results in the context of Deep RL (Finn et al., 2017; 2018).”
Formal Theory of Creativity, Fun, and Intrinsic Motivation (1990–2010) (2010, methodology): “To take the consumed computation time into account in a general, optimal way, one may use the recent Gödel machines [83], [86], [84], [98] instead.”
Artificial General Intelligence: Concept, State of the Art, and Future Prospects (2009, background): “Juergen Schmidhuber’s Goedel Machine (Schmidhuber, 2006) operates differently in detail, but the concept is similar.”
Artificial Intelligence as a Positive and Negative Factor in Global Risk (2006, background): “(Though see Schmidhuber (2007) for a related notion.)”