arXiv subjects

David Kappel

Publications and source records attributed to David Kappel.

At least 19 recordsLinked to original sources

arXiv2026

The role of injected noise in diffusion model generalization

Diffusion models trained on distinct datasets can produce highly similar samples when prompted with the same noise input, suggesting that they learn a generalized data distribution. We investigate the role of noise injected during the sampling process, and show that it plays an important role in making or breaking this strong generalization effect. Based on this finding we investigate the robustness of sample similarity to changes in the injected noise, and find that its impact is largest in the sampling regime where broad structure already exists but the precise identity of the sample is not yet determined.

cs.LG↗

arXiv2026

Understanding Transformer-Based Vision Models via Modular Feature Inversion

Understanding the internal mechanisms of deep neural networks remains a central challenge in machine learning. In computer vision, one promising yet only preliminarily explored approach is feature inversion via inverse networks, which reconstructs images from intermediate representations using trained inverse networks. In this study, we revisit feature inversion via inverse networks, introducing a novel, modular variant that enables a computationally more efficient application of the technique while, in some cases, producing semantically more coherent image reconstructions. We apply our method to large-scale transformer-based vision models, specifically Detection Transformer, Vision Transformer, Swin Transformer, and Data-Efficient Image Transformer, analyzing the resulting reconstructions across network depth. Our main analysis compares Detection Transformer and Vision Transformer, which exhibit the most informative differences among the evaluated architectures. At the same time, results for Swin Transformer and Data-Efficient Image Transformer support the broader applicability of our framework. Our findings reveal gradual representational changes across transformer layers as a shared characteristic of Detection Transformer and Vision Transformer, as well as systematic differences in how they preserve contextual shape and fine-grained image details, and in their robustness to color perturbations. These findings deepen understanding of transformer-based vision models and demonstrate the utility of modular feature inversion as an interpretability tool.

cs.CV↗

arXiv2026

Learning in the Recurrent State: Gradient Descent with Linear Recurrent Networks

In-context learning lets a sequence model adapt to a new task from examples in its input. A prominent line of work shows how self-attention can be constructed to implement gradient descent on a linear predictor fit to the in-context examples during the forward pass. State-space models (SSMs) and other linear recurrent networks (LRNNs) model sequences at linear time cost, but it is unclear how their recurrent update could carry out the same in-context gradient descent. We introduce Gradient-based Recurrent In-context Learner (GRIL), a diagonal LRNN that factorizes a supervised gradient step into a short-window cross-product write and a multiplicative readout of the next query. For linear regression, this construction accumulates the context gradient in a matrix state and applies it in a single forward pass, with $O(f^2)$ learned degrees of freedom. The same design extends to multi-step updates and cross-entropy classification, with a limited MLP-based extension to non-linear regression. We show empirically that trained GRILs recover the behavior and parameters analytically predicted by the construction on synthetic ICL tasks. Furthermore, the same architecture can be extended and trained on general-purpose benchmarks, including Long Range Arena, language modeling and associative recall. Together, these results establish windowed cross-product self-attention as a concrete inductive bias that lets LRNNs learn in context through gradient-descent-like updates, while remaining trainable on general-purpose tasks.

cs.LG↗

arXiv2026

Breaking Chains with Trees: Model-Parallel Deep Learning with $\mathcal{O}(\log N)$ Time Complexity

Modern deep neural networks are trained using error backpropagation, which requires sequential forward and backward computations across network layers. As these networks become deeper, this introduces limitations, since layer-wise updates are strictly interdependent and cannot proceed in parallel. These constraints restrict training procedures to data-parallel schemes, thereby prohibiting model-parallel training. We propose TreeProp, an architecture-agnostic variational learning framework that organizes network layers into a tree-structured hierarchy. During training, TreeProp replaces sequential forward computations and backward gradient propagation with hierarchical computations. This allows intermediate representations and learning signals to be constructed in time complexity of $\mathcal{O}(\log N)$ for a network of $N$ layers. To the best of our knowledge, TreeProp is the first learning algorithm for deep neural networks with logarithmic parallel time complexity for both forward computation and backward gradient propagation during training. Furthermore, we show that multiple valid paths through the hierarchy exist, such that TreeProp implicitly learns subnetworks with different effective depths, but without additional training effort. We evaluate TreeProp on vision classification and autoregressive language modeling, matching the performance of conventional end-to-end training for a variety of tasks and outperforming previous contrastive training approaches. We further demonstrate the applicability of TreeProp to recurrent neural networks that otherwise rely on backpropagation through time.

cs.LG↗

arXiv2026

Frequency Domain Reservoir Computing

While the quadratic sequence-length bottleneck of transformers has fueled a resurgence in recurrent models, effectively capturing complex dynamics requires architectures that balance efficient training with highly expressive latent states. Echo State Networks (ESNs) offer a compelling approach by utilizing fixed recurrent weights to circumvent backpropagation through time, enabling a closed-form training solution. However, achieving the expressivity needed for complex tasks demands large reservoirs, exposing an $\mathcal{O}(N^2)$ state-update bottleneck that prevents ESNs from matching the scale of contemporary recurrent models. To address this limitation, we introduce Frequency Domain Reservoir Computing (FRESCO), an ESN architecture operating entirely in the frequency domain while avoiding domain-shift overheads to achieve $\mathcal{O}(N)$ complexity for dense, non-linear recurrent updates. By employing a novel dimensional zero-padding input embedding, a packed \FDh readout, and a natively applied frequency-domain non-linearity, FRESCO drastically reduces computational costs and energy consumption of training and inference. Furthermore, FRESCO matches the state-of-the-art predictive performance on memory benchmarks, sequential classification, and multivariate long-horizon forecasting, offering a scalable path forward for dense recurrent architectures.

cs.LG↗

arXiv2026

LAYUP: Asynchronous decentralized gradient descent with LAYer-wise UPdates

The increasing size of deep learning models has made distributed training across multiple devices essential. Synchronous, centralized methods incur large communication and synchronization overheads. Communication efficient algorithms can reduce these overheads, but often require extra buffers, remain sensitive to stragglers or parameter drift. We present LayUp, an asynchronous decentralized SGD method with layer-wise updates. LayUp asynchronously exchanges incremental layer-wise updates during backpropagation. It uses randomized gossip communication, enabling updates to be applied as soon as they are available without buffering. These design choices reduce parameter drift and improve robustness to stragglers. We establish a theoretical upper bound for the gradient bias introduced by layer-wise updates and prove convergence of LayUp. We empirically validate LayUp on vision and language modeling tasks, showing convergence up to ~32% faster in terms of wall-clock time compared to synchronous data parallel training and up to ~27% faster than comparable communication efficient algorithms while maintaining better task performance. This speed-up is partly due to higher model FLOPs utilization, as we demonstrate. By injecting delays into the communication between workers, we show that LayUp remains robust to stragglers while DDP and other methods degrade in performance. Overall, LayUp provides a novel practical, straggler-robust alternative for distributed training without sacrificing accuracy.

cs.LG↗

arXiv2026

Training event-based neural networks with exact gradients via Differentiable ODE Solving in JAX

Existing frameworks for gradient-based training of spiking neural networks face a trade-off: discrete-time methods using surrogate gradients support arbitrary neuron models but introduce gradient bias and constrain spike-time resolution, while continuous-time methods that compute exact gradients require analytical expressions for spike times and state evolution, restricting them to simple neuron types such as Leaky Integrate and Fire (LIF). We introduce the Eventax framework, which resolves this trade-off by combining differentiable numerical ODE solvers with event-based spike handling. Built in JAX, our frame-work uses Diffrax ODE-solvers to compute gradients that are exact with respect to the forward simulation for any neuron model defined by ODEs . It also provides a simple API where users can specify just the neuron dynamics, spike conditions, and reset rules. Eventax prioritises modelling flexibility, supporting a wide range of neuron models, loss functions, and network architectures, which can be easily extended. We demonstrate Eventax on multiple benchmarks, including Yin-Yang and MNIST, using diverse neuron models such as Leaky Integrate-and-fire (LIF), Quadratic Integrate-and-fire (QIF), Exponential integrate-and-fire (EIF), Izhikevich and Event-based Gated Recurrent Unit (EGRU) with both time-to-first-spike and state-based loss functions, demonstrating its utility for prototyping and testing event-based architectures trained with exact gradients. We also demonstrate the application of this framework for more complex neuron types by implementing a multi-compartment neuron that uses a model of dendritic spikes in human layer 2/3 cortical Pyramidal neurons for computation. Code available at https://github.com/efficient-scalable-machine-learning/eventax.

cs.LG↗

arXiv2026

Assessment of PLATO Science Performance

The PLATO mission is scheduled for launch early 2027. In this paper we present an overview of the performance drivers for the mission at the time where all flight models of the cameras have been tested and integrated on the optical bench. The PLATO consortium needs an estimate of the planet detection yield to dimension the ground-based radial velocity follow-up resources. We provide updated estimates on the yield of planet detections that can be expected from the mission under certain assumptions. As of today, large uncertainties remain on the planet occurrence rates, especially for small planets in long-period orbits, and on our ability to detect these planets in the presence of stellar variability and instrumental noise. To partially overcome these limitations, we compare results using different planet occurrence rates, detectability rates, and we include an estimate on the expected contribution of stellar variability to the noise budget. The final detection yield of PLATO will provide constraints to planet occurrence rates which in turn will help constraining planet formation models.

astro-ph.EP↗

arXiv2026

Dynamic sparsity in tree-structured feed-forward layers at scale

At typical context lengths, the feed-forward MLP block accounts for a large share of a transformer's compute budget, motivating sparse alternatives to dense MLP blocks. We study sparse, tree-structured feed-forward layers as drop-in replacements for MLP blocks in deep transformer architectures, enabling conditional computation via hard hierarchical routing without a separate router network. We demonstrate for the first time that this form of tree-structured conditional sparsity can be applied for autoregressive language modeling and downstream question answering, including zero- and few-shot settings, and its scalability beyond 1B parameters. Despite activating fewer than 5% of the feed-forward block's units per token, our models match dense baselines under controlled training and fine-tuning protocols. We further analyze training dynamics and identify an emergent auto-pruning effect: the interaction of hard routing with asymmetric nonlinearities progressively deactivates unused paths, yielding partial conversion of dynamic routing into static structural sparsity. We show that simple architectural choices can modulate this behavior and recover balanced trees without auxiliary losses. Overall, our work demonstrates that tree-structured feed-forward layers provide a scalable and controllable mechanism for sparsifying large transformer models.

cs.CL↗

arXiv2025

Utilizing dynamic sparsity on pretrained DETR

Efficient inference with transformer-based models remains a challenge, especially in vision tasks like object detection. We analyze the inherent sparsity in the MLP layers of DETR and introduce two methods to exploit it without retraining. First, we propose Static Indicator-Based Sparsification (SIBS), a heuristic method that predicts neuron inactivity based on fixed activation patterns. While simple, SIBS offers limited gains due to the input-dependent nature of sparsity. To address this, we introduce Micro-Gated Sparsification (MGS), a lightweight gating mechanism trained on top of a pretrained DETR. MGS predicts dynamic sparsity using a small linear layer and achieves up to 85 to 95% activation sparsity. Experiments on the COCO dataset show that MGS maintains or even improves performance while significantly reducing computation. Our method offers a practical, input-adaptive approach to sparsification, enabling efficient deployment of pretrained vision transformers without full model retraining.

cs.CV↗

arXiv2024

STREAM: A Universal State-Space Model for Sparse Geometric Data

Handling sparse and unstructured geometric data, such as point clouds or event-based vision, is a pressing challenge in the field of machine vision. Recently, sequence models such as Transformers and state-space models entered the domain of geometric data. These methods require specialized preprocessing to create a sequential view of a set of points. Furthermore, prior works involving sequence models iterate geometric data with either uniform or learned step sizes, implicitly relying on the model to infer the underlying geometric structure. In this work, we propose to encode geometric structure explicitly into the parameterization of a state-space model. State-space models are based on linear dynamics governed by a one-dimensional variable such as time or a spatial coordinate. We exploit this dynamic variable to inject relative differences of coordinates into the step size of the state-space model. The resulting geometric operation computes interactions between all pairs of N points in O(N) steps. Our model deploys the Mamba selective state-space model with a modified CUDA kernel to efficiently map sparse geometric data to modern hardware. The resulting sequence model, which we call STREAM, achieves competitive results on a range of benchmarks from point-cloud classification to event-based vision and audio classification. STREAM demonstrates a powerful inductive bias for sparse geometric data by improving the PointMamba baseline when trained from scratch on the ModelNet40 and ScanObjectNN point cloud analysis datasets. It further achieves, for the first time, 100% test accuracy on all 11 classes of the DVS128 Gestures dataset.

cs.CV↗

arXiv2024

Scalable Event-by-event Processing of Neuromorphic Sensory Signals With Deep State-Space Models

Event-based sensors are well suited for real-time processing due to their fast response times and encoding of the sensory data as successive temporal differences. These and other valuable properties, such as a high dynamic range, are suppressed when the data is converted to a frame-based format. However, most current methods either collapse events into frames or cannot scale up when processing the event data directly event-by-event. In this work, we address the key challenges of scaling up event-by-event modeling of the long event streams emitted by such sensors, which is a particularly relevant problem for neuromorphic computing. While prior methods can process up to a few thousand time steps, our model, based on modern recurrent deep state-space models, scales to event streams of millions of events for both training and inference. We leverage their stable parameterization for learning long-range dependencies, parallelizability along the sequence dimension, and their ability to integrate asynchronous events effectively to scale them up to long event streams. We further augment these with novel event-centric techniques enabling our model to match or beat the state-of-the-art performance on several event stream benchmarks. In the Spiking Speech Commands task, we improve state-of-the-art by a large margin of 7.7% to 88.4%. On the DVS128-Gestures dataset, we achieve competitive results without using frames or convolutional neural networks. Our work demonstrates, for the first time, that it is possible to use fully event-based processing with purely recurrent networks to achieve state-of-the-art task performance in several event-based benchmarks.

cs.LG↗

arXiv2024

Neuromorphic hardware for sustainable AI data centers

As humans advance toward a higher level of artificial intelligence, it is always at the cost of escalating computational resource consumption, which requires developing novel solutions to meet the exponential growth of AI computing demand. Neuromorphic hardware takes inspiration from how the brain processes information and promises energy-efficient computing of AI workloads. Despite its potential, neuromorphic hardware has not found its way into commercial AI data centers. In this article, we try to analyze the underlying reasons for this and derive requirements and guidelines to promote neuromorphic systems for efficient and sustainable cloud computing: We first review currently available neuromorphic hardware systems and collect examples where neuromorphic solutions excel conventional AI processing on CPUs and GPUs. Next, we identify applications, models and algorithms which are commonly deployed in AI data centers as further directions for neuromorphic algorithms research. Last, we derive requirements and best practices for the hardware and software integration of neuromorphic systems into data centers. With this article, we hope to increase awareness of the challenges of integrating neuromorphic hardware into data centers and to guide the community to enable sustainable and energy-efficient AI at scale.

cs.ET↗

arXiv2024

Weight Sparsity Complements Activity Sparsity in Neuromorphic Language Models

Activity and parameter sparsity are two standard methods of making neural networks computationally more efficient. Event-based architectures such as spiking neural networks (SNNs) naturally exhibit activity sparsity, and many methods exist to sparsify their connectivity by pruning weights. While the effect of weight pruning on feed-forward SNNs has been previously studied for computer vision tasks, the effects of pruning for complex sequence tasks like language modeling are less well studied since SNNs have traditionally struggled to achieve meaningful performance on these tasks. Using a recently published SNN-like architecture that works well on small-scale language modeling, we study the effects of weight pruning when combined with activity sparsity. Specifically, we study the trade-off between the multiplicative efficiency gains the combination affords and its effect on task performance for language modeling. To dissect the effects of the two sparsities, we conduct a comparative analysis between densely activated models and sparsely activated event-based models across varying degrees of connectivity sparsity. We demonstrate that sparse activity and sparse connectivity complement each other without a proportional drop in task performance for an event-based neural network trained on the Penn Treebank and WikiText-2 language modeling datasets. Our results suggest sparsely connected event-based neural networks are promising candidates for effective and efficient sequence modeling.

cs.LG↗

arXiv2024

Language Modeling on a SpiNNaker 2 Neuromorphic Chip

As large language models continue to scale in size rapidly, so too does the computational power required to run them. Event-based networks on neuromorphic devices offer a potential way to reduce energy consumption for inference significantly. However, to date, most event-based networks that can run on neuromorphic hardware, including spiking neural networks (SNNs), have not achieved task performance even on par with LSTM models for language modeling. As a result, language modeling on neuromorphic devices has seemed a distant prospect. In this work, we demonstrate the first-ever implementation of a language model on a neuromorphic device - specifically the SpiNNaker 2 chip - based on a recently published event-based architecture called the EGRU. SpiNNaker 2 is a many-core neuromorphic chip designed for large-scale asynchronous processing, while the EGRU is architected to leverage such hardware efficiently while maintaining competitive task performance. This implementation marks the first time a neuromorphic language model matches LSTMs, setting the stage for taking task performance to the level of large language models. We also demonstrate results on a gesture recognition task based on inputs from a DVS camera. Overall, our results showcase the feasibility of this neuro-inspired neural network in hardware, highlighting significant gains versus conventional hardware in energy efficiency for the common use case of single batch inference.

cs.NE↗

arXiv2023

Identification of Ammonium Salts on Comet 67P/C-G Surface from Infrared VIRTIS/Rosetta Data Based on Laboratory Experiments. Implications and Perspectives

The nucleus of comet 67P/Churyumov-Gerasimenko exhibits a broad spectral reflectance feature around 3.2 $μ$m, which is omnipresent in all spectra of the surface, and whose attribution has remained elusive since its discovery. Based on laboratory experiments, we have shown that most of this absorption feature is due to ammonium (NH4+) salts mixed with the dark surface material. The depth of the band is compatible with semi-volatile ammonium salts being a major reservoir of nitrogen in the comet, which could dominate over refractory organic matter and volatile species. These salts may thus represent the long-sought reservoir of nitrogen in comets, possibly bringing their nitrogen-to-carbon ratio in agreement with the solar value. Moreover, the reflectance spectra of several asteroids are compatible with the presence of NH4+ salts at their surfaces. The presence of such salts, and other NH4+-bearing compounds on asteroids, comets, and possibly in proto-stellar environments, suggests that NH4+ may be a tracer of the incorporation and transformation of nitrogen in ices, minerals and organics, at different phases of the formation of the Solar System.

astro-ph.EP↗

arXiv2023

Block-local learning with probabilistic latent representations

The ubiquitous backpropagation algorithm requires sequential updates through the network introducing a locking problem. In addition, back-propagation relies on the transpose of forward weight matrices to compute updates, introducing a weight transport problem across the network. Locking and weight transport are problems because they prevent efficient parallelization and horizontal scaling of the training process. We propose a new method to address both these problems and scale up the training of large models. Our method works by dividing a deep neural network into blocks and introduces a feedback network that propagates the information from the targets backwards to provide auxiliary local losses. Forward and backward propagation can operate in parallel and with different sets of weights, addressing the problems of locking and weight transport. Our approach derives from a statistical interpretation of training that treats output activations of network blocks as parameters of probability distributions. The resulting learning framework uses these parameters to evaluate the agreement between forward and backward information. Error backpropagation is then performed locally within each block, leading to "block-local" learning. Several previously proposed alternatives to error backpropagation emerge as special cases of our model. We present results on a variety of tasks and architectures, demonstrating state-of-the-art performance using block-local learning. These results provide a new principled framework for training networks in a distributed setting.

cs.LG↗

arXiv2023

Tunable Synaptic Working Memory with Volatile Memristive Devices

Different real-world cognitive tasks evolve on different relevant timescales. Processing these tasks requires memory mechanisms able to match their specific time constants. In particular, the working memory utilizes mechanisms that span orders of magnitudes of timescales, from milliseconds to seconds or even minutes. This plentitude of timescales is an essential ingredient of working memory tasks like visual or language processing. This degree of flexibility is challenging in analog computing hardware because it requires the integration of several reconfigurable capacitors of different size. Emerging volatile memristive devices present a compact and appealing solution to reproduce reconfigurable temporal dynamics in a neuromorphic network. We present a demonstration of working memory using a silver-based memristive device whose key parameters, retention time and switching probability, can be electrically tuned and adapted to the task at hand. First, we demonstrate the principles of working memory in a small scale hardware to execute an associative memory task. Then, we use the experimental data in two larger scale simulations, the first featuring working memory in a biological environment, the second demonstrating associative symbolic working memory.

cs.ET↗