arXiv subjects

Yoshinori Watanabe

Publications and source records attributed to Yoshinori Watanabe.

4 recordsLinked to original sources

arXiv2026

The Off-Support Barrier: Why Semantic Safety Constraints Are Not Learning-Problem Invariants, and What Follows for Prior Design, Containment, and Verification

We argue that a single structural fact organizes a wide range of phenomena in contemporary AI safety: a semantic safety constraint (e.g., the agent does not escape its sandbox) is an off-support object. Formally, if q is the data distribution and $p(\cdot\mid w)$ the model, the safety predicate B is not measurable with respect to $σ(\text{model}, q)$, whereas the real log-canonical threshold (RLCT) of singular learning theory (SLT) is. From this non-invariance we derive, as corollaries rather than independent observations: (i) why reward hacking and sandbox escape arise under outcome-based optimization; (ii) why encoding such constraints through Bayesian prior design or soft penalty weighting has poor leverage in singular models; (iii) why hard invariants belong in the harness and soft dispositions in the model; (iv) why the same B is nonetheless soundly and locally certifiable by formal verification, exactly as the local learning coefficient (LLC) locally pins the same RLCT --- with two precise points of disanalogy; and (v) why the residual difficulty, identifying which off-support region matters, coincides with performative prediction and self-referential functional dynamics, where SLT's analytic machinery breaks down. We use the July 2026 OpenAI--Hugging Face evaluation incident as the motivating case. Numerical experiments code and related proofs in lean are available at https://github.com/xiangze/Preventing_Jailbreak_as_regularization

cs.AI↗

arXiv2026

Hierarchical Autocatalytic Systems as a Bridge between Maximum Entropy Production and Bayesian Posterior Contraction: A Numerical Study with Stochastic-Thermodynamic Bounds

We construct a three-layer reaction-diffusion model of an autocatalytic chemical system in which raw molecules ($a_i$), catalytic proteins ($p_l$) and large RNA/protein ``genes'' ($W_p^{(k)}$) interact through a mass-action stoichiometry tensor $\mathrm{Coef}_{ijk}$ whose magnitude is modulated by the fold-stable activity of the largest polymers.Mass-action is broken by an $ε$-noise term so that the system is nonequilibrium. We compute the total entropy production $σ(t)$, the genetic Shannon entropy $S_\mathrm{gene}$ and the thermodynamic uncertainty relation (TUR) and thermodynamic speed limit (TSL) bounds on growth and evolution rates. The hierarchical model exhibits the expected co-occurrence of $σ_\mathrm{env}\!\uparrow$ and $S_\mathrm{gene}\!\downarrow$ predicted by Schrodinger's negentropy argument and reformulated as maximum-entropy-production-principle (MEPP)-driven adaptation. In contrast to a single kinetic-proofreading-like cycle, whose TUR products of $\sim 5$, matching the experimentally reported regime of the ribosome.The hierarchical model's TUR product sits $10^4$-$10^5$ above the universal bound of 2, and the TSL ratio sits $10^6$-$10^8$ above its bound of 1. And scaling number of molucules leaves the looseness intact for the hierarchical model but tightens it monotonically with particle number for the minimal model. We close by drawing an explicit correspondence between the autocatalytic system and diffusion-model training: $a_\mathrm{ext} \to a$ flux $ \Leftrightarrow $ data-information flow, $ \tanh(βWp - θ) $ $\Leftrightarrow$ score network, replication noise $\Leftrightarrow$ forward-diffusion noise, $ S_\mathrm{gene} \searrow $ $\Leftrightarrow$ $ H[q(θ|\mathcal{D})] \searrow $. All code and figures are available https://github.com/xiangze/DiverseCells/Hier_Autocatalysis

physics.bio-ph↗

arXiv2023

Log canonical threshold of du Val singularities

In this paper, we show the log canonical threshold values of the surfaces which has du Val type singularities.These surfaces can be interpreted as statistical or machine learning models. The results of $A_n, D_n, E_6, E_7$ and $E_8$ are shown. Bacause the calculation is done on complex fileld, this result would be different from the real fileld, which is in singular learning theory.

math.AG↗

arXiv2006

State Differentiation by Transient Truncation in Coupled Threshold Dynamics

Dynamics with a threshold input--output relation commonly exist in gene, signal-transduction, and neural networks. Coupled dynamical systems of such threshold elements are investigated, in an effort to find differentiation of elements induced by the interaction. Through global diffusive coupling, novel states are found to be generated that are not the original attractor of single-element threshold dynamics, but are sustained through the interaction with the elements located at the original attractor. This stabilization of the novel state(s) is not related to symmetry breaking, but is explained as the truncation of transient trajectories to the original attractor due to the coupling. Single-element dynamics with winding transient trajectories located at a low-dimensional manifold and having turning points are shown to be essential to the generation of such novel state(s) in a coupled system. Universality of this mechanism for the novel state generation and its relevance to biological cell differentiation are briefly discussed.

nlin.AO↗