argmax.ai · Machine Learning Research Lab, Volkswagen Group, Munich

about
The Machine Learning Research Lab in Volkswagen Group IT focuses on fundamental machine learning research for probabilistic inference, time series modelling, efficient exploration, and control. Our work starts from probability theory, through neural networks, to applications inside and outside our lab. This website is to inform you about our work through publications, blog posts, and other public output.

Latent Matters

Learning Deep State-Space Models

Inferring the state-space representation of a pendulum

10toGO: Virtual thinkathon for Sustainable Development Goals

Or: what we do when we don't do science.

AI for Sustainable Development Goals

Variational State-Space Models for Localisation and Dense 3D Mapping in 6 DoF

Scaling generative spatial models to the real world

Reconstruction of the "main square" of our office space with our method. Colour, depth and uncertainty.

Less Suboptimal Learning and Control in Variational POMPDs

How to fix model-based RL by doing the obvious.

Model learning fails under strong partial-observability.

A Tale of Gaps

Inference Suboptimalities in Amortised Variational Inference

A catastrophic example of the newly discovered conditioning gap.

Continual Learning with Bayesian Neural Networks for Non-Stationary Data

Adapting to changing environments

Online variational Bayes with Bayesian forgetting (left) and without adaptation (right).

Estimating Fingertip Forces, Torques, and Object Curvatures from Fingernail Images

picking up and placing

Layerwise learning for quantum neural networks

Solving vanishing gradients in quantum neural networks

Incrementally training layers solves vanishing gradients in quantum neural networks

Learning to Fly

Deep Model-Based Reinforcement Learning in the real world

A self-built drone controlled onboard by a learnt policy optimised in a learnt simulation

Learning Flat Manifold of VAEs

How to create latent spaces without nasty curvatures

Flat Manifold VAE

Approximate Bayesian inference in spatial environments

How to reason about space and moving agents

A mobile agent can localise itself in an unknown maze using DVBF-LM.

Learning Hierarchical Priors in VAEs

A constrained optimisation approach

Graph-based interpolation of human motion

How to Learn Functions on Sets with Neural Networks

And how to choose your aggregation

The basic Deep Sets architecture for set functions: embed, aggregate, process.

Approximate Geodesics for Deep Generative Models

How to efficiently find the shortest path in latent space

The graph of the fashion MNIST dataset in a 2D latent space, along with the magnification factor.

Network Architecture Optimisation

The Bayesian way

Deep Variational Bayes Filter

DVBF: filter to learn what to filter

Learnt latent representation of a swinging pendulum

About

Machine Learning Research Lab, Volkswagen Group

The front view of the lab building

about
The Machine Learning Research Lab in Volkswagen Group IT focuses on fundamental machine learning research for probabilistic inference, time series modelling, efficient exploration, and control. Our work starts from probability theory, through neural networks, to applications inside and outside our lab. This website is to inform you about our work through publications, blog posts, and other public output.

Learn More