arXiv subjects

Fabian A. Mikulasch

Publications and source records attributed to Fabian A. Mikulasch.

4 recordsLinked to original sources

arXiv2026

Learning Commute-Time-Preserving World Models for Planning

World models allow agents to plan in latent space by choosing a sequence of actions that most reduces the distance to a given goal state. Thus, planning can benefit from latent representations whose distances mirror commute-times in the environment. The spectral embedding space of the graph Laplacian provides such a representation, if it obeys a specific eigenvalue-dependent scaling. Unfortunately, instantiating the graph Laplacian is intractable in large, continuous environments. Self-supervised learning offers a natural route to such commute-time-preserving embeddings at scale. However, here we show that existing methods, which commonly encourage isotropic representations to prevent representational collapse, tend to degrade the "correct" eigenvalue-dependent scaling, leading to an inaccurate representation of commute times. To address this problem, we introduce Commute-Time-Preserving World Models (CTWMs), combining a latent displacement predictor and a log-determinant regularizer that prevents collapse, which provably recover the correctly scaled Laplacian representation under reversible deterministic dynamics and at the predictor's fixed point. In numerical simulations, CTWM matches or outperforms LeWM, a task-agnostic baseline, on several complex, continuous goal-reaching benchmarks, while using half the parameters.

cs.LG↗

arXiv2026

Predictive Self-Supervised Learning Provably Identifies Stochastic Signals under Nuisance

Self-supervised learning (SSL) by predicting in latent space, without generating the input data itself, learns highly abstract, useful representations. Intuitively, this success is often attributed to its ability to discard nuisance information that is irrelevant to prediction. However, this poses a conundrum: both stochastic variation in a prediction-relevant latent signal and true nuisance make observations partly unpredictable; how could they be distinguished? Surprisingly, we prove that common SSL methods can achieve exactly this, by implicitly instantiating a latent-variable model with stochastic dynamics and observation-private nuisance. We trace their ability to recover the stochastic signal to two complementary principles: Predictive mutual information maximization ensures that representations retain the information needed for prediction, while latent distribution matching constrains how this information is encoded, thereby making the retained signal identifiable. We confirm this identifiability result in simulations for Gaussian predictors, which recover the true signal up to an affine transformation even in dynamic, nuisance-laden environments.

cs.LG↗

arXiv2022

Dendritic predictive coding: A theory of cortical computation with spiking neurons

Top-down feedback in cortex is critical for guiding sensory processing, which has prominently been formalized in the theory of hierarchical predictive coding (hPC). However, experimental evidence for error units, which are central to the theory, is inconclusive, and it remains unclear how hPC can be implemented with spiking neurons. To address this, we connect hPC to existing work on efficient coding in balanced networks with lateral inhibition, and predictive computation at apical dendrites. Together, this work points to an efficient implementation of hPC with spiking neurons, where prediction errors are computed not in separate units, but locally in dendritic compartments. The implied model shows a remarkable correspondence to experimentally observed cortical connectivity patterns, plasticity and dynamics, and at the same time can explain hallmarks of predictive processing, such as mismatch responses, in cortex. We thus propose dendritic predictive coding as one of the main organizational principles of cortex.

q-bio.NC↗

arXiv2018

Sparse Coding Predicts Optic Flow Specificities of Zebrafish Pretectal Neurons

Zebrafish pretectal neurons exhibit specificities for large-field optic flow patterns associated with rotatory or translatory body motion. We investigate the hypothesis that these specificities reflect the input statistics of natural optic flow. Realistic motion sequences were generated using computer graphics simulating self-motion in an underwater scene. Local retinal motion was estimated with a motion detector and encoded in four populations of directionally tuned retinal ganglion cells, represented as two signed input variables. This activity was then used as input into one of two learning networks: a sparse coding network (competitive learning) and backpropagation network (supervised learning). Both simulations develop specificities for optic flow which are comparable to those found in a neurophysiological study (Kubo et al. 2014), and relative frequencies of the various neuronal responses are best modeled by the sparse coding approach. We conclude that the optic flow neurons in the zebrafish pretectum do reflect the optic flow statistics. The predicted vectorial receptive fields show typical optic flow fields but also "Gabor" and dipole-shaped patterns that likely reflect difference fields needed for reconstruction by linear superposition.

q-bio.NC↗