Funded research projects
In reverse chronological order:
POST-DIGITAL is
a European H2020 Marie Curie Innovative Training Network (2020-2023)
whose 15 partner institutions will provide academic and applied training opportunities to a cohort of 15
early stage researchers (ESRs) in neuromorphic computational
technologies and their applications. A strong industrial presence in
the network will provide our ESRs with the experience of practical
applications and solutions beyond traditional digital methods. MINDS
is leading the mathematical theory branch of research in this
consortium.
MeM-scales (Memory technologies with multi-scale time constants for neuromorphic architectures) is a
European H2020 ICT project (2020-2022) with a consortium of 9
academic and industrial partners. The mission is to build ultra-low
power embedded and pervasive computing systems which realize core
aspects of cognitive computing, in particular integrating information
across several memory timescales. MINDS is leading the mathematical
theory branch of research in this consortium.
Projects before 2020 (when the MINDS group was hosted in Jacobs University Bremen):
NeuRAM3
(Neural Computing Architectures in Advanced Monolithic
3D-VLSI Nano-Technologies) was a European H2020 Project
(2016-2019) which developed a novel neuromorphic VLSI chip
architecture and fabrication technology. Coordinated by the
Commisariat à l’énergie atomique et aux énergies alternatives (CEA)
and uniting 9 partners (among them gloablly leading chip
manufacturers), the consortium aimed at a low power
consumption (reduction by a factor of 50 compared to current
technology) and flexible configurability. MINDS
developed novel neural learning architectures for robustness against noise, parameter drift,
and low numerical accuracy.
AMARSi (Adaptive Modular Architectures for Rich Motor Skills)
was a European Collaborative Project (IST-248311, 2010-2014) which aimed
at a qualitative jump in robotic motor skills toward biological
richness. Coordinated from the COR Lab
at the University of Bielefeld, Amarsi joined 10 partners from
robotics, neural computation and computational neuroscience, the motion
sciences, and cognitive science. MINDS contributed models of
neurocontrollers and led the architectures workpackage.
ORGANIC (Self-Organized Recurrent Neural Learning for Language Processing)
was a European FP7 project (IST-231267, 2009-2012) whose
mission was to establish neurodynamical architectures as viable
alternative to statistical methods for speech and handwriting
recognition. Coordinated by MINDS, it
comprised six European research groups from reservoir computing,
cognitive neuroscience, and speech and handwriting technology. MINDS
contributed layered neural learning architectures based on Echo
State Networks.
Observable operator networks. Funded by the DFG (contract JA 1210/5-1, 2009-2012). A framework for unifying a number of predictive state based theories independently proposed in different disciplines was worked out; the foundations for spectral optimization of OOM learning algorithms were laid; and OOM learning algorithms were extended to data with missing values (PhD thesis of Michael Thon).
Quadratic observable operator models. Funded by the DFG (contract JA 1210/1-1&2, 2005-2009), this project developed a number of statistically and computationally efficient learning algorithms for observable operator models (OOMs), and established a theory of quadratic OOMs. These describe stochastic processes by linear operators in an intriguing analogy to the formalism of quantum mechanics. The results were published in three articles in Neural Computation. Principal researcher: Mingjie Zhao.
Industrial
handwriting recognition solutions. In a lively
collaboration with PLANET intelligent
systems GmbH, MINDS was regularly supported by PLANET through
stipends for PhD students and student projects concerned with Echo
State Networks algorithms for handwriting recognition.