arXiv subjects

Kevin Max

Publications and source records attributed to Kevin Max.

9 recordsLinked to original sources

arXiv2026

Dendritic structure enables powerful plasticity

Over the past decades, it has become increasingly clear that the complex morphology of cortical neurons is more than just a quirk of evolution, and that dendritic compartments serve as computational elements in their own right, rather than just providing connections between nerve cell bodies. While most computational studies discuss the enhanced representational capabilities of multi-compartment models as compared to point neurons, we focus here on the implications of neuronal morphology for synaptic plasticity. We argue that the ability of single neurons to simultaneously encode multiple pieces of information gives synapses local access to more than just the classical Hebbian pre- and postsynaptic terms, and with much greater specificity and reaction speed than permitted by other globally modulated factors. Based on a comparative review of recent dendritic learning models, we show how such neuronal compartmentalization can provide synapses with the means for calculating various forms of error signals, which in turn give rise to powerful real-time and fully local instantiations of deep learning through gradient descent. Implemented within cortical microcircuits capable of propagating and manipulating these errors, compartmentalized neurons thus ultimately enable the learning of far more complex tasks than are achievable by globally modulated Hebbian plasticity alone.

q-bio.NC↗

arXiv2025

Synthetic Biology meets Neuromorphic Computing: Towards a bio-inspired Olfactory Perception System

In this study, we explore how the combination of synthetic biology, neuroscience modeling, and neuromorphic electronic systems offers a new approach to creating an artificial system that mimics the natural sense of smell. We argue that a co-design approach offers significant advantages in replicating the complex dynamics of odor sensing and processing. We propose a hybrid system of synthetic sensory neurons that provides three key features: (a) receptor-gated ion channels, (b) interface between synthetic biology and semiconductors and (c) event-based encoding and computing based on spiking networks. Our approach is validated using simulation-based modeling of the complete sensing and processing pipeline. This research seeks to develop a platform for ultra-sensitive, specific, and energy-efficient odor detection, with potential implications for environmental monitoring, medical diagnostics, and security.

cs.NE↗

arXiv2025

Spike-based alignment learning solves the weight transport problem

In both machine learning and in computational neuroscience, plasticity in functional neural networks is frequently expressed as gradient descent on a cost. Often, this imposes symmetry constraints that are difficult to reconcile with local computation, as is required for biological networks or neuromorphic hardware. For example, wake-sleep learning in networks characterized by Boltzmann distributions assumes symmetric connectivity. Similarly, the error backpropagation algorithm is notoriously plagued by the weight transport problem between the representation and the error stream. Existing solutions such as feedback alignment circumvent the problem by deferring to the robustness of these algorithms to weight asymmetry. However, they scale poorly with network size and depth. We introduce spike-based alignment learning (SAL), a complementary learning rule for spiking neural networks, which uses spike timing statistics to extract and correct the asymmetry between effective reciprocal connections. Apart from being spike-based and fully local, our proposed mechanism takes advantage of noise. Based on an interplay between Hebbian and anti-Hebbian plasticity, synapses can thereby recover the true local gradient. This also alleviates discrepancies that arise from neuron and synapse variability -- an omnipresent property of physical neuronal networks. We demonstrate the efficacy of our mechanism using different spiking network models. First, SAL can significantly improve convergence to the target distribution in probabilistic spiking networks versus Hebbian plasticity alone. Second, in neuronal hierarchies based on cortical microcircuits, SAL effectively aligns feedback weights to the forward pathway, thus allowing the backpropagation of correct feedback errors. Third, our approach enables competitive performance in deep networks using only local plasticity for weight transport.

q-bio.NC↗

arXiv2024

Backpropagation through space, time, and the brain

How physical networks of neurons, bound by spatio-temporal locality constraints, can perform efficient credit assignment, remains, to a large extent, an open question. In machine learning, the answer is almost universally given by the error backpropagation algorithm, through both space and time. However, this algorithm is well-known to rely on biologically implausible assumptions, in particular with respect to spatio-temporal (non-)locality. Alternative forward-propagation models such as real-time recurrent learning only partially solve the locality problem, but only at the cost of scaling, due to prohibitive storage requirements. We introduce Generalized Latent Equilibrium (GLE), a computational framework for fully local spatio-temporal credit assignment in physical, dynamical networks of neurons. We start by defining an energy based on neuron-local mismatches, from which we derive both neuronal dynamics via stationarity and parameter dynamics via gradient descent. The resulting dynamics can be interpreted as a real-time, biologically plausible approximation of backpropagation through space and time in deep cortical networks with continuous-time neuronal dynamics and continuously active, local synaptic plasticity. In particular, GLE exploits the morphology of dendritic trees to enable more complex information storage and processing in single neurons, as well as the ability of biological neurons to phase-shift their output rate with respect to their membrane potential, which is essential in both directions of information propagation. For the forward computation, it enables the mapping of time-continuous inputs to neuronal space, effectively performing a spatio-temporal convolution. For the backward computation, it permits the temporal inversion of feedback signals, which consequently approximate the adjoint variables necessary for useful parameter updates.

q-bio.NC↗

arXiv2022

Learning efficient backprojections across cortical hierarchies in real time

Models of sensory processing and learning in the cortex need to efficiently assign credit to synapses in all areas. In deep learning, a known solution is error backpropagation, which however requires biologically implausible weight transport from feed-forward to feedback paths. We introduce Phaseless Alignment Learning (PAL), a bio-plausible method to learn efficient feedback weights in layered cortical hierarchies. This is achieved by exploiting the noise naturally found in biophysical systems as an additional carrier of information. In our dynamical system, all weights are learned simultaneously with always-on plasticity and using only information locally available to the synapses. Our method is completely phase-free (no forward and backward passes or phased learning) and allows for efficient error propagation across multi-layer cortical hierarchies, while maintaining biologically plausible signal transport and learning. Our method is applicable to a wide class of models and improves on previously known biologically plausible ways of credit assignment: compared to random synaptic feedback, it can solve complex tasks with less neurons and learn more useful latent representations. We demonstrate this on various classification tasks using a cortical microcircuit model with prospective coding.

q-bio.NC↗

arXiv2020

Searching for Elusive Dark Sectors with Terrestrial and Celestial Observations

We consider the possible existence of a SM-neutral and light dark sector coupled to the visible sector through irrelevant portal interactions. Scenarios of this kind are motivated by dark matter and arise in various extensions of the Standard Model. We characterize the dark dynamics in terms of one ultraviolet scale $\Lambda_\text{UV}$, at which the exchange of heavy mediator fields generates the portal operators, and by one infrared scale $\Lambda_\text{IR}$, setting the mass gap. At energies $\Lambda_\text{IR} \ll E \ll \Lambda_\text{UV}$ the dark sector behaves like a conformal field theory and its phenomenology can be studied model independently. We derive the constraints set on this scenario by high- and low-energy laboratory experiments and by astrophysical observations. Our results are conservative and serve as a minimum requirement that must be fulfilled by the broad class of models satisfying our assumptions, of which we give several examples. The experimental constraints are derived in a manner consistent with the validity of the effective field theory used to define the portal interactions. We find that high-energy colliders give the strongest bounds and exclude UV scales up to a few TeV, but only in specific ranges of the IR scale. The picture emerging from current searches can be taken as a starting point to design a future experimental strategy with broader sensitivity.

hep-ph↗

arXiv2020

Probing alternative cosmologies through the inverse distance ladder

We study the implications of a combined analysis of cosmic standard candles and standard rulers on the viability of cosmological models beyond the cosmological concordance model. To this end, we employ well-established data in the form of the joint light-curve analysis supernova compilation, baryon acoustic oscillations, and cosmic microwave background data on the one hand, and a recently proposed set of quasars as objects of known brightness on the other hand. The advantage of including the latter is that they extend the local distance measures to redshifts which have previously been out of reach and we investigate how this allows one to test cosmologies beyond $\Lambda$CDM. While there exist various studies on parametric extensions of $\Lambda$CDM, we present here a comparative study of both parametric and fundamental extensions of the standard cosmology. In order to keep the scope of this manuscript contained, we focus on two particular modifications: One is the consistent theory of two interacting spin-2 objects, so-called bigravity, and the other conformal gravity, a theory of gravity that has no knowledge of fundamental length scales. The former of the two constitutes a veritable extension of General Relativity, given that it adds to the metric tensor of gravity a second dynamical tensor field. The resulting dynamics have been proposed as a self-accelerating cosmology. Conformal gravity on the other hand is a much more drastic change of the underlying gravitational theory. Its ignorance towards fundamental length scales offers a completely different approach to the origin of late time acceleration. In this sense, both models offer - in one way or another - an explanation for the dark energy problem. We perform a combined cosmological fit which provides strong constraints on these extensions. We also briefly comment on the implications of the long-standing $H_0$-tension.

astro-ph.CO↗

arXiv2017

Decoherence of Gravitational Wave Oscillations in Bigravity

Following up on our recent study, we consider the regime of graviton masses and gravitational wave propagation distances at which decoherence of the wave packets plays a major role for phenomenology. This regime is of particular interest, as it can lead to very striking phenomena of echo events in the gravitational waves coming from coalescence events. The power of the experimental search in this case lies in the fact that it becomes sensitive to a large range of graviton masses, while not relying on a specific production mechanism. We are thus able to place new relevant limits on the parameter space of the graviton mixing angle.

gr-qc↗

arXiv2017

Gravitational Wave Oscillations in Bigravity

We derive consistent equations for gravitational wave oscillations in bigravity. In this framework a second dynamical tensor field is introduced in addition to General Relativity and coupled such that one massless and one massive linear combination arise. Only one of the two tensors is the physical metric coupling to matter, and thus the basis in which gravitational waves propagate is different from the basis where the wave is produced and detected. Therefore, one should expect -- in analogy to neutrino oscillations -- to observe an oscillatory behavior. We show how this behavior arises explicitly, discuss phenomenological implications and present new limits on the graviton parameter space in bigravity.

gr-qc↗