Most of the studies so far have employed datasets for their analysis with only limited number of neurons. The landscape now changes with the rapid advances in new optical methods. The neuroscience partners of the consortium have access to i) Mesoscopic in vivo 2-photon calcium imaging that affords exceptionally large-field of view, significantly increasing the yield of the imaging, ii) SLM (spatial light modulation) based optogenetic probing of networks, and iii) in vivo imaging of the PFC in freely-moving animals with the miniscope technology. These are combined with in vitro and in vivo patch clamping giving us the opportunity to dissect neural circuity in detail.

Rich datasets can be acquired essentially simultaneously across different cortical layers at ~1-20Hz for long periods of time, providing us with a unique opportunity to analyze them in order to understand how functional networks of neurons behave in the biological brain, with particular focus on how they derive their computational properties and organization during i) stimulus encoding, ii) feed-forward information transfer between areas, and iii) learning.
It will be also the first time that network interactions across different areas at a global, whole-brain level, will be cross correlated with detailed measurements collected at a fine spatio-temporal scale, with single cell resolution. For the whole-brain level analysis, high field functional magnetic resonance imaging (fMRI).

It will be also the first time that network interactions across different areas at a global, whole-brain level, will be cross correlated with detailed measurements collected at a fine spatio-temporal scale, with single cell resolution. For the whole-brain level analysis, high field functional magnetic resonance imaging (fMRI) and resting state functional magnetic resonance imaging (rsfMRI) datasets will be employed.

FORTH (ICS) has extensive experience in analyzing complex large-scale real-world networks using powerful tools for correlation, state change identification, and network comparisons. It will extend and apply such tools taking into consideration the underlying characteristics of the neuronal activity patterns and collected measurements (noise, biases, large scale, dynamic characteristics, types of neurons, state changes). For example, it will be the first time to apply graph similarity algorithms in such large, of fine spatio-termporal scale measurements as the ones that will be collected here. To enhance the effectiveness of the analysis, which will be shared with the scientific community, neuronsXnets plan to develop scalable platforms for efficient analysis and data sharing.

The expected findings will have broad applicability and are not restricted in biology. Potential advances include developing new computational paradigms, including devising algorithms for optimizing the performance of current deep machine learning algorithms as well as improving current imaging acquisition and preprocessing techniques. Despite the remarkable progress, today’s artificial systems are still not able to compete with biological ones in tasks that involve processing of sensory data acquired in real-time, in complex and uncertain settings. One of the reasons is that neural computation in biological systems is very different from the way today’s computers operate. Neuromorphic circuits are a class of hybrid analog/digital circuits that implement hardware models of biological systems. It has been argued that these types of circuits can be used to develop a new generation of computing technologies based on the organizing principles of the biological nervous system. The styles of computation used in neuromorphic circuits are fundamentally different from those used by conventional computers. As the biological systems they model, they process information using energy-efficient asynchronous, event-driven, methods and most importantly are adaptive and fault-tolerant.

These characteristics offer an attractive alternative to conventional computing strategies, especially if one considers the advantages and potential problems of future advanced VLSI fabrication processes. Innovative systems optimized for higher levels of cognition, exceeding the performance of the current deep machine-learning algorithms, can employ these findings to make robust inferences about the world from limited noisy data. UZH, IBM, and Bruker, three of the partners of the consortium, have been performing pioneering research in neuromorphic computing and imaging techniques, respectively. NeuronXnets will take advantage of the multi-disciplinarity and research excellence to make innovative contributions in these areas.