Bert de Vries | BIASlab

AIM-TT

Mon, 01 Dec 2025 00:00:00 +0000

AiMTT aims to cultivate a highly skilled and diverse AI talent pool equipped to address the opportunities and challenges of AI in mobility, transport, and logistics. By combining real-world case studies with knowledge development, this initiative fosters deep expertise in the field.

Mobility, transport, and logistics face a multitude of challenges: traffic congestion, livability concerns, conflicts between user, operator, and public interests, space constraints, and safety risks during large-scale events. These challenges are further complicated by their deep interconnections, making them particularly difficult to resolve.

While such complexities can be overwhelming for the human mind, they offer ideal use cases for artificial intelligence (AI). AI can process vast amounts of data in real time, provide accurate network assessments, calculate impacts in future scenarios, and optimize interventions. It also enhances our understanding of human behavior and the mobility system as a whole.

Given these advantages, leveraging AI to address mobility challenges is a logical next step. Through AiMTT, we embrace a “learning by doing” approach to develop responsible, AI-driven solutions.

Vision and Ambition

AiMTT stands for AI Learning Initiative for Multi-modal Traffic and Transportation. It operates under the umbrella of AIC4NL, an organization dedicated to the responsible development and application of AI in the Netherlands. Ensuring responsible AI development is essential, as concerns about fairness, inclusivity, privacy, and human oversight continue to grow. Will AI-generated outcomes be equitable? Can privacy be safeguarded? How do we ensure that humans remain in control?

AiMTT aims to address these critical questions by fostering a collaborative learning community that brings together experts from academia, industry, and government. Project partners will build, test, and refine AI applications, with ethical considerations—such as fairness, privacy, and human autonomy—at the forefront.

Learning Process

AiMTT’s approach is grounded in practical application. AI solutions will be developed through seven real-world use cases, each designed to create tangible tools that can be directly implemented. Equally important is the learning process itself: identifying best practices, analyzing challenges, and refining AI applications based on real-world insights.

To support this, the project will offer workshops, training programs, and co-creation sessions—ensuring continuous knowledge exchange and improvement.

Through AiMTT, we are shaping the future of urban mobility by responsibly integrating AI to create smarter, safer, and more efficient transportation systems.

A Message Passing Realization of Expected Free Energy Minimization

Fri, 17 Oct 2025 00:00:00 +0000

A Factor Graph Approach to Variational Sparse Gaussian Processes

Wed, 02 Jul 2025 09:00:00 +0200

FEP-Lab

Sun, 01 Jun 2025 00:00:00 +0000

The FEPlab (Free Energy Principle Laboratory) is a collaboration between Eindhoven University of Technology (TU/e) and GN Hearing. The mission of the lab is to ameliorate the participation of hearing-impaired people in formal and informal social settings. The lab will focus its research on transferring a leading physics/neuroscience-based theory about computation in the brain, the Free Energy Principle (FEP), to practical use in human-centered agents such as hearing devices and VR technology.

GN Hearing, which is a globally leading hearing aid manufacturer with a strong research team (of about 20 persons) in Eindhoven, and the TU/e have already been collaborating for many years in BIASlab, which is a research team at the Electrical Engineering department at TU/e. This collaboration has produced theoretical foundations for synthetic FEP-based AI agents. FEPlab has been set up in 2022 and is expected to run until mid-2027. During this time, the partners will continue to develop these FEP agents into a technology that is ready for deployment in the professional hearing device industry.

FEPlab focuses on two Sustainable Development Goals: Goal 3, Good Health and Well-being, and Goal 5, Promote sustained, inclusive, and sustainable economic growth, full and productive employment, and decent work for all. Untreated hearing loss in the elderly increases the risk of developing dementia and Alzheimer’s disease (Ralli et al., 2019) as well as emotional and physical problems (Ciorba et al., 2012). Therefore, this research neatly ties into SDG3 Target 1: reducing premature mortality from non-communicable diseases. Moreover, hearing loss negatively impacts work participation (Svinndal et al., 2018). Hence, this research also ties into SDG8 Target 1: achieve higher levels of economic productivity through technology upgrading and innovation.

The lab comprises experts from different fields of expertise such as Audiology, Autonomous Agents & Robotics, Decision Making, and Machine Learning to tackle the complex multidisciplinary challenges at hand. Socially aware AI and explainable AI are especially important in the lab’s research since the technology needs to be aware of the social context in which it is operating and be able to provide justification for its decisions and actions in a manner that is understandable by humans to ensure its safe use.

Ralli, Massimo, et al. “Hearing loss and Alzheimer’s disease: A Review.” The international tinnitus journal 23.2 (2019): 79-85.
Ciorba, Andrea, et al. “The impact of hearing loss on the quality of life of elderly adults.” Clinical interventions in aging 7 (2012): 159.
Svinndal, Elisabeth Vigrestad, et al. “Hearing loss and work participation: a cross-sectional study in Norway.” International journal of audiology 57.9 (2018): 646-656.

The lab is part of the Innovation Center for Artificial Intelligence

ExponentialFamilyManifolds.jl: Representing exponential families as Riemannian manifolds

Mon, 14 Apr 2025 16:37:31 +0200

GraphPPL.jl: A Probabilistic Programming Language for Graphical Models

Tue, 22 Oct 2024 09:00:00 +0200

Improved Depth Estimation of Bayesian Neural Networks

Mon, 14 Oct 2024 09:00:00 +0200

Riemannian Black Box Variational Inference

Mon, 14 Oct 2024 09:00:00 +0200

Message Passing-based Bayesian Control of a Cart-Pole System

Wed, 11 Sep 2024 09:00:00 +0200

Online Structure Learning with Dirichlet Processes through Message Passing

Wed, 11 Sep 2024 09:00:00 +0200

Q-conjugate Message Passing for Efficient Bayesian Inference

Wed, 11 Sep 2024 09:00:00 +0200

Reactive Environments for Active Inference Agents with RxEnvironments.jl

Wed, 11 Sep 2024 09:00:00 +0200

Multi-Agent Trajectory Planning with NUV Priors

Wed, 10 Jul 2024 00:00:00 +0000

Bayesian inference of collision avoidance intent during ship encounters

Tue, 02 Apr 2024 09:00:00 +0200

Principled Pruning of Bayesian Neural Networks through Variational Free Energy Minimization

Wed, 29 Nov 2023 00:00:00 +0000

Gaussian Process Amplitude Demodulation by Message-Passing

Sun, 17 Sep 2023 00:00:00 +0000

Efficient Bayesian Inference by Conjugate-computation Variational Message Passing

Sat, 16 Sep 2023 18:07:00 +0200

Toward Design of Synthetic Active Inference Agents by Mere Mortals

Thu, 14 Sep 2023 00:00:00 +0000

Automating model comparison in factor graphs

Sat, 29 Jul 2023 00:00:00 +0000

Realising Synthetic Active Inference Agents, Part II: Variational Message Updates

Wed, 12 Jul 2023 09:00:00 +0200

Realising Synthetic Active Inference Agents, Part I: Epistemic Objectives and Graphical Specification Language

Fri, 16 Jun 2023 09:00:00 +0200

Reactive Message Passing for Scalable Bayesian Inference

Sat, 27 May 2023 09:00:00 +0200

RxInfer: A Julia package for reactive real-time Bayesian inference

Thu, 20 Apr 2023 09:00:00 +0200

Efficient Model Evidence Computation in Tree-structured Factor Graphs

Fri, 04 Nov 2022 00:00:00 +0000

Online Single-Microphone Source Separation using Non-Linear Autoregressive Models

Mon, 19 Sep 2022 00:00:00 +0000

Hybrid Inference with Invertible Neural Networks in Factor Graphs

Fri, 02 Sep 2022 09:00:00 +0200

Gaussian Process-based Amortization of Variational Message Passing Update Rules

Fri, 02 Sep 2022 00:00:00 +0000

Message Passing-based Inference in Switching Autoregressive Models

Fri, 02 Sep 2022 00:00:00 +0000

Adaptive importance sampling message passing

Sun, 26 Jun 2022 00:00:00 +0000

Probabilistic programming with stochastic variational message passing

Wed, 22 Jun 2022 09:00:00 +0200

Adaptive Optimizer Design for Constrained Variational Inference

Wed, 01 Jun 2022 00:00:00 +0000

ReactiveMP.jl: A Julia package for reactive variational Bayesian inference

Sun, 01 May 2022 00:00:00 +0000

Active Inference and Epistemic Value in Graphical Models

Wed, 06 Apr 2022 09:00:00 +0200

AIDA: An active inference-Based design agent for audio processing algorithms

Mon, 07 Mar 2022 00:00:00 +0000

ReactiveMP.jl: A Julia package for reactive message passing-based Bayesian inference

Sat, 29 Jan 2022 00:00:00 +0000

On Epistemics in Expected Free Energy for Linear Gaussian State Space Models

Wed, 24 Nov 2021 10:40:00 +0200

Message Passing-Based Inference in the Gamma Mixture Model

Mon, 25 Oct 2021 00:00:00 +0000

A Bayesian Modeling Approach to Situated Design of Personalized Soundscaping Algorithms

Thu, 14 Oct 2021 11:37:31 +0200

Auto-AR

Fri, 01 Oct 2021 00:00:00 +0000

Automated Situated Design of Augmented Hearing Reality Algorithms

Bayesian pure-tone audiometry through active learning under informed priors

Fri, 13 Aug 2021 07:41:37 +0100

Variational Log-Power Spectral Tracking for Acoustic Signals

Tue, 13 Jul 2021 00:00:00 +0000

The Switching Hierarchical Gaussian Filter

Mon, 12 Jul 2021 00:00:00 +0000

Extended Variational Message Passing for Automated Approximate Bayesian Inference

Sat, 26 Jun 2021 16:37:31 +0200

Variational Message Passing and Local Constraint Manipulation in Factor Graphs

Thu, 24 Jun 2021 16:37:31 +0200

Message Passing-Based Inference for Time-Varying Autoregressive Models

Fri, 28 May 2021 16:37:31 +0200

BayesBrain

Thu, 01 Apr 2021 00:00:00 +0000

Computation in biological brain tissue consumes several orders of magnitude less power than silicon-based systems. Motivated by this fact, this project aims to develop the world’s first hybrid neuro-in-silico Artificial Intelligence (AI) computer, introducing a fundamentally new paradigm of AI computing. In this high-risk high-gain project, we will combine an in-silico Bayesian control agent (BCA) with neural tissue hosted by a microfluidic Brain-on-Chip (BoC) that together form a hybrid learning system capable of solving real-world AI problems.

All computation and communication inside and between the BCA and BoC will be governed by the Free Energy Principle, which is both the leading neuroscientific theory for describing biological neuronal processes and supports a variational Bayesian machine learning interpretation. We will start by developing a pure silicon-based BCA that learns to balance an inverted pendulum, implemented by free energy minimization on a factor graph. Next, we will replace successively larger parts of the factor graph with biological neural circuits of a microfluidic multi-compartment BoC device. The biological network will be trained by electrical stimulation orchestrated by the synthetic Bayesian agent. For the communication between these two units, we will design and realize a novel communication protocol making use of existing software being applied in readout and event sorting for Calcium imaging and multi-electrode array data, such as MEAViewer, CALIMA, NetCal and SpikeHunter. By upscaling the number of replaced sub-circuits, we aim to provide a proof-of-concept and to lay the basis for ultra-low power hybrid brain-on-chip AI computing.

This position is supported by the Exploratory Multidisciplinary AI Research Program of the Eindhoven Artificial Intelligence Systems Institute.

Learning Where to Park

Tue, 15 Sep 2020 16:37:31 +0200

A Worked Example of Fokker-Planck based Active Inference

Mon, 14 Sep 2020 10:40:00 +0200

Online Message Passing-based Inference in the Hierarchical Gaussian Filter

Mon, 22 Jun 2020 00:00:00 +0000

Online Variational Message Passing in Hierarchical Autoregressive Models

Mon, 22 Jun 2020 00:00:00 +0000

Bayesian joint state and parameter tracking in autoregressive models

Mon, 08 Jun 2020 00:00:00 +0000

Real-time Audio Processing for Hearing Aids using a Model-Based Bayesian Inference Framework

Tue, 26 May 2020 09:00:00 +0200

Approximate Inference by Kullback-Leibler Tensor Belief Propagation

Wed, 06 May 2020 18:30:00 +0200

BATMAN: Bayesian target modelling for active inference

Wed, 06 May 2020 18:30:00 +0200

Reparameterization Gradient Message Passing

Mon, 02 Sep 2019 09:00:00 +0200

Online Variational Message Passing in Autoregressive Models

Tue, 28 May 2019 18:07:00 +0200

Robust Bayesian beamforming for sources at different distances with applications in urban monitoring

Fri, 17 May 2019 18:07:00 +0200

Simulating Active Inference Processes by Message Passing

Thu, 28 Mar 2019 16:37:31 +0200

A Factor Graph Approach to Automated Design of Bayesian Signal Processing Algorithms

Tue, 01 Jan 2019 14:50:37 +0100

ForneyLab.jl: Fast and flexible automated inference through message passing in Julia

Fri, 05 Oct 2018 08:07:00 +0200

Online Variational Message Passing in the Hierarchical Gaussian Filter

Sun, 23 Sep 2018 18:07:00 +0200

ForneyLab: A Toolbox for Biologically Plausible Free Energy Minimization in Dynamic Neural Models

Sun, 23 Sep 2018 13:42:00 +0200

Acoustic scene classification from few examples

Sun, 09 Sep 2018 09:07:00 +0200

Robust Expectation Propagation in Factor Graphs Involving Both Continuous and Binary Variables

Thu, 06 Sep 2018 14:07:00 +0200

ForneyLab.jl: a Julia Toolbox for Factor Graph-based Probabilistic Programming

Wed, 08 Aug 2018 13:36:00 +0200

A Probabilistic Modeling Approach to One-Shot Gesture Recognition

Fri, 06 Jul 2018 16:37:31 +0200

K-shot learning of acoustic context

Fri, 08 Dec 2017 16:37:31 +0200

A parametric approach to Bayesian optimization with pairwise comparisons

Sun, 03 Dec 2017 16:37:31 +0200

A factor graph description of deep temporal active inference

Wed, 18 Oct 2017 00:00:00 +0000

Variational Stabilized Linear Forgetting in State-Space Models

Sat, 30 Sep 2017 13:10:00 +0200

A probabilistic modeling approach to hearing loss compensation

Tue, 11 Jul 2017 13:19:00 +0200

The graphical brain: Belief propagation and active inference

Fri, 07 Jul 2017 00:00:00 +0000

A Gaussian process mixture prior for hearing loss modeling

Fri, 09 Jun 2017 16:37:31 +0200

An In-situ Trainable Gesture Classifier

Fri, 09 Jun 2017 16:37:31 +0200

Probabilistic Inference-based Reinforcement Learning

Fri, 09 Jun 2017 16:37:31 +0200

ZERO-AAS

Thu, 01 Jun 2017 00:00:00 +0000

An NWO-TTW Perspectief program called ZERO, organized around 5 projects:

Parking sensors, such that cars can find a free parking slot autonomously.
Monitoring of traffic patterns using advanced audio beam processing.
Autonomous roadside monitoring with video.
Ultra-low power transponders for vulnerable traffic users.
Dependable autonomous computing platforms supporting mobile traffic users.

We were involved in P2, dubbed Autonomous Acoustic Systems (AAS). AAS can be found in various shapes and sizes, ranging from city wide acoustic monitoring systems to hearing aids worn by an individual. To deliver an improved quality and user experience, future generations of these systems should use adaptive signal processing algorithms, while staying within a stringent energy budget for autonomous operation. The AAS project uses two of these systems as driver cases to develop a novel programming paradigm and accompanying ultra-low power implementation platform for a wide range of autonomous acoustic systems.

Industrial partners: Resound, Sorama and Altran

CoHear

Wed, 01 Feb 2017 00:00:00 +0000

Hearing loss is a very serious health condition that has been associated with early dementia and cognitive decline. Still, hearing aid market penetration is quite low, in particular for the large population that is afflicted with ‘mild-to-moderate’ hearing impairment. This is mainly due to two reasons: Stigma (association with old age) and hearing aids (HA) sound quality. The recent commercial introduction of fashionable ‘hearables’ will likely alleviate the stigma issue. Recent advances in machine learning and cloud computing open new avenues for attacking the sound quality issue for hearing aids. In this project, we intend to develop a (crowd-based) collaborative design approach to improving the sound quality issues for the mild-to-moderately hearing-impaired population.

We expect to build a working prototype for collaboratively designed hearing algorithms that can be applied to a new class of ‘smart hearing devices’ with high appeal to the mild-to-moderately hearing-impaired patient. As an additional benefit, we hope that our technology will ease the transition from hearables to professional hearing aid technology for the moderate-to-profound hearing-impaired population.

A Probabilistic Modeling Approach to Hearing Loss Compensation

Thu, 10 Nov 2016 16:37:31 +0200

A Bayesian binary classification approach to pure tone audiometry

Thu, 10 Sep 2015 16:37:31 +0200

A Probabilistic Approach To Hearing Loss Compensation

Wed, 10 Sep 2014 09:00:00 +0200