theory

"In the case of independent test data, we obtain a dimension-free rate of order O(n^{-1/2} ) on the n-sample generalization error, whereas without independence assumption, we derive a bound of order O(n^{-1 / ( d_{rm in}+d_{rm out} )} ), where d_{rm in}, d_{rm out} denote input and output dimensions."

"Hi, I am currently making the jump to ML from theoretical physics."

"We investigate whether suppressing mind-attribution tendencies degrades intimately related socio-cognitive abilities such as Theory of Mind (ToM)."

"The difference is only one: how to measure distance."

"Rather than constantly poring over theory "what attention is, what normalization is" it’s better to train the model yourself and look for anomalies."

"How do you go from an intuitive idea to a theoretical justification?"

"GPT-4o demonstrated high accuracy and strong robustness, performing comparably to humans even in the most challenging conditions."

"In 2022, Bodnar et al. presented "Neural Sheaf Diffusion" at NeurIPS, introducing the 200-year-old theory of sheaves into GNNs and resolving both oversmoothing and heterophily in a single stroke."

"The message was impressive that the difference between whether or not to set limits before or after the challenge greatly affects the scope of possibility."

"In AI-related X(Twitter) disputes, Japanese-language debates showed a higher proportion of ad hominem attacks compared to English-language debates, while English-language debates showed a higher rate of process disclosure — these trends are interpretable within the frameworks of Edward T. Hall's context theory and Stella Ting-Toomey's Face Negotiation Theory."

"Mutual information is a measure of how much knowing one variable reduces uncertainty about another."

"By employing the mathematical treasure of “Sheaf Cohomology,” we can transform these contradictions into something to be measured."

"Having been cautious about AI, Knuth acknowledged, "I seem to have to revise my opinion about Generative AI.""

"The reason was that he had solved the unsolved problem he had been working on for several weeks, with Anthropic's AI model Claude Opus 4.6 in just one hour."

"This work develops a dynamical theory of sequential reasoning in Hopfield networks."

"When an event x occurs with probability p(x), its self-information (surprisal) is: I(x) = -log2 p(x) [bit]"

"Koah, which embeds sponsored ads directly into AI chatbots and aims to build an AdSense for AI, raised a $20.5M Series A led by Theory Ventures"

"The article addresses the limits of LLMs - subjection to probability, forgetting of context, inability to reverse knowledge, inability to plan, insincerity of Chain of Thought, and inevitability of Hallucination - from the single perspective of 'absence of integration'."

"I’m comfortable with the math side of things and can understand the concepts, but I’m struggling with the practical coding part."

"Rebuilding "Shining Force Neo" with Modern Technology A Hack-and-Slash Design Spec Using Graph Theory and AI"

"We show that contextuality is not a peculiarity of quantum mechanics, but an inevitable consequence of single-state reuse in classical probabilistic representations."

"I argue that AI literacy cannot be acquired through declarative knowledge alone, but requires guided exposure to AI failure modes, including the sycophantic validation and authority projection patterns identified by Sharma et al."

"Across 15 frontier models, agents choose socially beneficial actions in only 62% of cases, frequently leading to harmful outcomes."

"It is the first time I’ve seen AI solve a problem in my kind of theoretical physics that might not have been solvable by humans."

"some recent research by Anthropic has deeply much debunked it."

"Our analysis on iterated rock-paper-scissors reveals that frontier LLMs can be capable of deeper strategic behavior than humans."

"In the limit where the number of data samples $n$ and the dimension $d$ grow proportionally, we derive a closed system of low-dimensional and continuous-time equations and prove that it characterizes the asymptotic distribution of the SGF parameters."

""Everything fell into place naturally after that," says Chen, who worke"