Revisiting Convolutional Neural Networks from the Viewpoint of Kernel-Based Methods

Chat with Graph Search

Ask any question about EPFL courses, lectures, exercises, research, news, etc. or try the example questions below.

DISCLAIMER: The Graph Chatbot is not programmed to provide explicit or categorical answers to your questions. Rather, it transforms your questions into API requests that are distributed across the various IT services officially administered by EPFL. Its purpose is solely to collect and recommend relevant references to content that you can explore to help you answer your questions.

The ability of the brain to produce adaptive and dexterous movements relies on the seamless integration of sensory feedback into motor commands. Despite significant advances in motor neuroscience, critical gaps persist in understanding how proprioceptive a ...

Computational methods at the atomic scale based on machine learning have been a fruitful research field in the past decade by merging the advantages of classical and quantum mechanical approaches. While parts of the field are already reaching a mature stat ...

A recent line of work in high-dimensional statistics working under the Gaussian mixture hypothesis has led to a number of results in the context of empirical risk minimization, Bayesian uncertainty quantification, separation of kernel methods and neural ne ...

Recently, remarkable progress has been made in the application of machine learning (ML) techniques (e.g., neural networks) to transformer fault diagnosis. However, the diagnostic processes employed by these techniques often suffer from a lack of interpreta ...

Optimum groundwater management is important for addressing the global challenges of water scarcity, especially in arid regions. This study demonstrates the application of synthetic data combined with seismic methods to quantify water volumes in a controlle ...

Gradient-based learning in multi-layer neural networks displays a number of striking features. In particular, the decrease rate of empirical risk is non-monotone even after averaging over large batches. Long plateaus in which one observes barely any progre ...

Over the past years, implantable closed-loop neural interfaces have received increasing attention for their potential to enhance the understanding of the brain, treat neurological disorders, and provide unprecedented opportunities for individuals with para ...

Semantic segmentation with deep learning plays a crucial role in various fields, including civil engineering, particularly in tasks such as damage assessment and urban planning. This paper addresses the challenge of efficiently training deep learning model ...

In recent years, numerical simulations of hemodynamics have gained significant attention within the medical community, thanks to their ability of non-invasively estimating the blood flow conditions. However, high-fidelity simulations require extensive comp ...

Explainable Artificial Intelligence (XAI) plays a crucial role in enabling human understanding and trust in deep learning systems, often defined as determining which features are most important to a model's prediction. As models get larger, more ubiquitous ...

Recently there has been a surge of interest in understanding implicit regularization properties of iterative gradient-based optimization algorithms. In this paper, we study the statistical guarantees on the excess risk achieved by early-stopped unconstrain ...

This paper proposes a computational approach to form-find pin-jointed bar structures subjected to combinations of tension and compression forces. The generated equilibrium states can meet structural and geometrical constraints via gradient-based optimizati ...

We consider a class of spreading processes on networks, which generalize commonly used epidemic models such as the SIR model or the SIS model with a bounded number of reinfections. We analyze the related problem of inference of the dynamics based on its pa ...

Event-driven acquisition uses neural-network-based recognition of specific biological events to trigger switching between slow and fast super-resolution imaging, enriching the capture of interesting events with high spatiotemporal resolution. A common goal ...

Advances in soft sensors coupled with machine learning are enabling increasingly capable wearable systems. Since hand motion in particular can convey useful information for developing intuitive interfaces, glove-based systems can have a significant impact ...

This thesis is centered on questions coming from Machine Learning (ML) and Statistical Field Theory (SFT). In Machine Learning, we consider the subfield of Supervised Learning (SL), and in particular regression tasks where one tries to find a regressor tha ...

Inferences using Convolutional Neural Networks (CNNs) are resource and energy intensive. Therefore, their execution on highly constrained edge devices demands the careful co-optimization of algorithms and hardware. Addressing this challenge, in this paper ...

Recent research has shown that it is possible to find interpretable directions in the latent spaces of pre-trained Generative Adversarial Networks (GANs). These directions enable controllable image generation and support a wide range of semantic editing op ...

Revisiting Convolutional Neural Networks from the Viewpoint of Kernel-Based Methods

Chat with Graph Search

Reverse engineering primate sensorimotor control with machine learning

Mathematical Analysis and Optimization of Features for Atomistic Machine Learning

Universality laws for Gaussian mixtures in generalized linear models

Explainable Fault Diagnosis of Oil-Immersed Transformers: A Glass-Box Model

Uncertainty quantification in groundwater volume predictions from seismic data using neural networks

Learning Time-Scales in Two-Layers Neural Networks

Neurostimulation and Wireless Data Transmission Circuits for Implantable High-Density Neural Interfaces

Partial annotations in active learning for semantic segmentation

Reduced order models to address temporal complexity and geometrical variability in hemodynamics

The future of human-centric eXplainable Artificial Intelligence (XAI) is not post-hoc explanations

The statistical complexity of early-stopped mirror descent

Constrained Form-Finding of Tension-Compression Structures using Automatic Differentiation

Bayes-optimal inference for spreading processes on random networks

Event-driven acquisition for content-enriched microscopy

A Wearable Smart Glove and Its Application of Pose and Gesture Detection to Sign Language Classification

Complex Analysis meets Statistical Mechanics: Applications to Kernel Methods and Conformal Field Theory

A Flexible In-Memory Computing Architecture for Heterogeneously Quantized CNNs

LatentCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions