A5005914878- Neural dynamics and brain function
- Advanced Memory and Neural Computing
- Neural Networks and Applications
- Visual perception and processing mechanisms
- Neural Networks and Reservoir Computing
- Ferroelectric and Negative Capacitance Devices
- Neuroscience and Neural Engineering
- Auction Theory and Applications
- Model Reduction and Neural Networks
- Game Theory and Applications
- Face Recognition and Perception
- Healthcare Operations and Scheduling Optimization
- Time Series Analysis and Forecasting
- EEG and Brain-Computer Interfaces
- Neural and Behavioral Psychology Studies
- Reinforcement Learning in Robotics
- Visual Attention and Saliency Detection
- Stock Market Forecasting Methods
- Complex Systems and Time Series Analysis
- Gaussian Processes and Bayesian Inference
- Multi-Agent Systems and Negotiation
- Scheduling and Timetabling Solutions
- Consumer Market Behavior and Pricing
- Photoreceptor and optogenetics research
- stochastic dynamics and bifurcation
Centrum Wiskunde & Informatica
2016-2025
University of Amsterdam
2000-2024
Utrecht University
2024
University of Groningen
2020-2023
Netherlands Institute for Neuroscience
2021-2023
Rijksmuseum
2022
College of Western Idaho
2022
Amsterdam Neuroscience
2021
Centro Tecnológico de Investigación, Desarrollo e Innovación en tecnologías de la Información y las Comunicaciones (TIC)
2018
Vitenparken
2012
We present a method for conditional time series forecasting based on an adaptation of the recent deep convolutional WaveNet architecture. The proposed network contains stacks dilated convolutions that allow it to access broad range history when forecasting, ReLU activation function and conditioning is performed by applying multiple filters in parallel separate which allows fast processing data exploitation correlation structure between multivariate series. test analyze performance both...
We demonstrate that spiking neural networks encoding information in the timing of single spikes are capable computing and learning clusters from realistic data. show how a network based on spike-time coding Hebbian can successfully perform unsupervised clustering real-world data, we temporal synchrony multilayer induce hierarchical clustering. develop continuously valued data to obtain adjustable capacity precision with an efficient use neurons: input variables encoded population code by...
We present a method for conditional time series forecasting based on an adaptation of the recent deep convolutional WaveNet architecture. The proposed network contains stacks dilated convolutions that allow it to access broad range historical data when forecasting. It also uses rectified linear unit (ReLU) activation function, and conditioning is performed by applying multiple filters in parallel separate series, which allows fast processing exploitation correlation structure between...
The emergence of brain-inspired neuromorphic computing as a paradigm for edge AI is motivating the search high-performance and efficient spiking neural networks to run on this hardware. However, compared classical in deep learning, current lack competitive performance compelling areas. Here, sequential streaming tasks, we demonstrate how novel type adaptive recurrent network (SRNN) able achieve state-of-the-art other almost reach or exceed (RNNs) while exhibiting sparse activity. From this,...
This paper proposes a data-driven approach, by means of an Artificial Neural Network (ANN), to value financial options and calculate implied volatilities with the aim accelerating corresponding numerical methods. With ANNs being universal function approximators, this method trains optimized ANN on data set generated sophisticated model, runs trained as agent original solver in fast efficient way. We test approach three different types solvers, including analytic solution for Black-Scholes...
We present a novel reduced order model (ROM) approach for parameterized time-dependent PDEs based on modern learning. The ROM is suitable multi-query problems and nonintrusive. It divided into two distinct stages: nonlinear dimensionality reduction stage that handles the spatially distributed degrees of freedom convolutional autoencoders, time-stepping memory aware neural networks (NNs), specifically causal long short-term NNs. Strategies to ensure generalization stability are discussed. To...
Intelligence is our ability to learn appropriate responses new stimuli and situations. Neurons in association cortex are thought be essential for this ability. During learning these neurons become tuned relevant features start represent them with persistent activity during memory delays. This process not well understood. Here we develop a biologically plausible scheme that explains how trial-and-error induces neuronal selectivity working representations task-relevant information. We propose...
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease caused by aberrations in the genome. While several disease-causing variants have been identified, major part of heritability remains unexplained. ALS believed to complex genetic basis where non-additive combinations constitute disease, which cannot be picked up using linear models employed classical genotype-phenotype association studies. Deep learning on other hand highly promising for identifying such relations. We...
Artificial Neural Networks (ANNs) are bio-inspired models of neural computation that have proven highly effective. Still, ANNs lack a natural notion time, and units in exchange analog values frame-based manner, computationally energetically inefficient form communication. This contrasts sharply with biological neurons communicate sparingly efficiently using isomorphic binary spikes. While Spiking (SNNs) can be constructed by replacing the an ANN spiking (Cao et al., 2015; Diehl 2015) to...
Working memory is essential: it serves to guide intelligent behavior of humans and nonhuman primates when task-relevant stimuli are no longer present the senses. Moreover, complex tasks often require that multiple working representations can be flexibly independently maintained, prioritized, updated according changing task demands. Thus far, neural network models have been unable offer an integrative account how such control mechanisms acquired in a biologically plausible manner. Here, we...
The field of neuromorphic computing holds great promise in terms advancing efficiency and capabilities by following brain-inspired principles. However, the rich diversity techniques employed research has resulted a lack clear standards for benchmarking, hindering effective evaluation advantages strengths methods compared to traditional deep-learning-based methods. This paper presents collaborative effort, bringing together members from academia industry, define benchmarks computing:...
Biological neurons communicate with a sparing exchange of pulses - spikes. It is an open question how real spiking produce the kind powerful neural computation that possible deep artificial networks, using only so very few spikes to communicate. Building on recent insights in neuroscience, we present Adapting Spiking Neural Network (ASNN) based adaptive neurons. These efficiently encode information spike-trains form Asynchronous Pulsed Sigma-Delta coding while homeostatically optimizing...
Humans can quickly recognize objects in a dynamically changing world. This ability is showcased by the fact that observers succeed at recognizing rapidly image sequences, up to 13 ms/image. To date, mechanisms govern dynamic object recognition remain poorly understood. Here, we developed deep learning models for and compared different computational mechanisms, contrasting feedforward recurrent, single-image sequential processing as well forms of adaptation. We found only integrate images...
Localizing leakages in large water distribution systems is an important and ever-present problem. Due to the complexity originating from pipeline networks, too few sensors, noisy measurements, this a highly challenging problem solve. In work, we present methodology based on generative deep learning Bayesian inference for leak localization with uncertainty quantification. A model, utilizing neural serves as probabilistic surrogate model that replaces full equations, while at same time also...