-
Pure Data Foundation of Mathematics and Computing
Authors:
Saul Youssef
Abstract:
We propose an axiomatic foundation of mathematics based on the finite sequence as the foundational concept, rather than based on logic and set, as in set theory, or based on type as in dependent type theories. Finite sequences lead to a concept of pure data, which is used to represent all mathematical objects. As an axiomatic system, the foundation has only one axiom which defines what constitutes…
▽ More
We propose an axiomatic foundation of mathematics based on the finite sequence as the foundational concept, rather than based on logic and set, as in set theory, or based on type as in dependent type theories. Finite sequences lead to a concept of pure data, which is used to represent all mathematical objects. As an axiomatic system, the foundation has only one axiom which defines what constitutes a valid definition. Using the axiom, an internal true/false/undecided valued logic and an internal language are defined, making logic and language-related axioms unnecessary. Valid proof and valid computation are defined in terms of equality of pure data. An algebra of pure data leads to a rich theory of spaces and morphisms which play a role similar to the role of Category Theory in modern Mathematics. As applications, we explore Mathematical Machine Learning, the consistency of Mathematics and address paradoxes due to Godel, Berry, Curry and Yablo.
△ Less
Submitted 9 May, 2023;
originally announced May 2023.
-
Towards a Solution to Bongard Problems: A Causal Approach
Authors:
Salahedine Youssef,
Matej Zečević,
Devendra Singh Dhami,
Kristian Kersting
Abstract:
Even though AI has advanced rapidly in recent years displaying success in solving highly complex problems, the class of Bongard Problems (BPs) yet remain largely unsolved by modern ML techniques. In this paper, we propose a new approach in an attempt to not only solve BPs but also extract meaning out of learned representations. This includes the reformulation of the classical BP into a reinforceme…
▽ More
Even though AI has advanced rapidly in recent years displaying success in solving highly complex problems, the class of Bongard Problems (BPs) yet remain largely unsolved by modern ML techniques. In this paper, we propose a new approach in an attempt to not only solve BPs but also extract meaning out of learned representations. This includes the reformulation of the classical BP into a reinforcement learning (RL) setting which will allow the model to gain access to counterfactuals to guide its decisions but also explain its decisions. Since learning meaningful representations in BPs is an essential sub-problem, we further make use of contrastive learning for the extraction of low level features from pixel data. Several experiments have been conducted for analyzing the general BP-RL setup, feature extraction methods and using the best combination for the feature space analysis and its interpretation.
△ Less
Submitted 23 December, 2022; v1 submitted 14 June, 2022;
originally announced June 2022.
-
Minimal Economic Distributed Computing
Authors:
Saul Youssef,
John Brunelle,
John Huth,
David C. Parkes,
Margo Seltzer,
Jim Shank
Abstract:
In an ideal distributed computing infrastructure, users would be able to use diverse distributed computing resources in a simple coherent way, with guaranteed security and efficient use of shared resources in accordance with the wishes of the owners of the resources. Our strategy for approaching this ideal is to first find the simplest structure within which these goals can plausibly be achieved…
▽ More
In an ideal distributed computing infrastructure, users would be able to use diverse distributed computing resources in a simple coherent way, with guaranteed security and efficient use of shared resources in accordance with the wishes of the owners of the resources. Our strategy for approaching this ideal is to first find the simplest structure within which these goals can plausibly be achieved. This structure, we find, is given by a particular recursive distributive lattice freely constructed from a presumed partially ordered set of all data in the infrastructure. Minor syntactic adjustments to the resulting algebra yields a simple language resembling a UNIX shell, a concept of execution and an interprocess protocol. Persons, organizations and servers within the system express their interests explicitly via a hierarchical currency. The currency provides a common framework for treating authentication, access control and resource sharing as economic problems while also introducing a new dimension for improving the infrastructure over time by designing system components which compete with each other to earn the currency. We explain these results, discuss experience with an implementation called egg and point out areas where more research is needed.
△ Less
Submitted 26 February, 2009;
originally announced February 2009.
