Lyudmila Grigoryeva

Optimal nonlinear information processing capacity in delay-based reservoir computers.

Reservoir computing is a recently introduced brain-inspired machine learning paradigm capable of excellent performances in the processing of empirical data. We focus in a particular kind of time-delay based reservoir computers that have been physically implemented using optical and electronic systems and have shown unprecedented data processing rates. Reservoir computing is well-known for the ease of the associated training scheme but also for the problematic sensitivity of its performance to architecture parameters. This article addresses the reservoir design problem, which remains the biggest challenge in the applicability of this information processing scheme. More specifically, we use the information available regarding the optimal reservoir working regimes to construct a functional link between the reservoir parameters and its performance. This function is used to explore various properties of the device and to choose the optimal reservoir architecture, thus replacing the tedious and time consuming parameter scannings used so far in the literature.

Protocol Design Challenges in the Detection of Awareness in Aware Subjects Using EEG Signals

Recent studies have evidenced serious difficulties in detecting covert awareness with electroencephalography-based techniques both in unresponsive patients and in healthy control subjects. This work reproduces the protocol design in two recent mental imagery studies with a larger group comprising 20 healthy volunteers. The main goal is assessing if modifications in the signal extraction techniques, training-testing/cross-validation routines, and hypotheses evoked in the statistical analysis, can provide solutions to the serious difficulties documented in the literature. The lack of robustness in the results advises for further search of alternative protocols more suitable for machine learning classification and of better performing signal treatment techniques. Specific recommendations are made using the findings in this work.

Reservoir Computing: Information Processing of Stationary Signals

This paper extends the notion of information processing capacity for non-independent input signals in the context of reservoir computing (RC). The presence of input autocorrelation makes worthwhile the treatment of forecasting and filtering problems for which we explicitly compute this generalized capacity as a function of the reservoir parameter values using a streamlined model. The reservoir model leading to these developments is used to show that, whenever that approximation is valid, this computational paradigm satisfies the so called separation and fading memory properties that are usually associated with good information processing performances. We show that several standard memory, forecasting, and filtering problems that appear in the parametric stochastic time series context can be readily formulated and tackled via RC which, as we show, significantly outperforms standard techniques in some instances.

Echo state networks are universal

This paper shows that echo state networks are universal uniform approximants in the context of discrete-time fading memory filters with uniformly bounded inputs defined on negative infinite times. This result guarantees that any fading memory input/output system in discrete time can be realized as a simple finite-dimensional neural network-type state-space model with a static linear readout map. This approximation is valid for infinite time intervals. The proof of this statement is based on fundamental results, also presented in this work, about the topological nature of the fading memory property and about reservoir computing systems generated by continuous reservoir maps.

Time-Delay Reservoir Computers and High-Speed Information Processing Capacity

The aim of this presentation is to show how various ideas coming from the nonlinear stability theory of functional differential systems, stochastic modeling, and machine learning, can be put together in order to create an approximating model that explains the working mechanisms behind a certain type of reservoir computers. Reservoir computing is a recently introduced brain-inspired machine learning paradigm capable of excellent performances in the processing of empirical data. We focus on time-delay based reservoir computers that have been physically implemented using optical and electronic systems and have shown unprecedented data processing rates. Reservoir computing is well-known for the ease of the associated training scheme but also for the problematic sensitivity of its performance to architecture parameters. The reservoir design problem is addressed, which remains the biggest challenge in the applicability of this information processing scheme. Our results use the information available regarding the optimal reservoir working regimes in order to construct a functional link between the reservoir parameters and its performance. This function is used to explore various properties of the device and to choose the optimal reservoir architecture, thus replacing the tedious and time consuming parameter scannings used so far in the literature.

Bedside Evaluation of the Functional Organization of the Auditory Cortex in Patients with Disorders of Consciousness

To measure the level of residual cognitive function in patients with disorders of consciousness, the use of electrophysiological and neuroimaging protocols of increasing complexity is recommended. This work presents an EEG-based method capable of assessing at an individual level the integrity of the auditory cortex at the bedside of patients and can be seen as the first cortical stage of this hierarchical approach. The method is based on two features: first, the possibility of automatically detecting the presence of a N100 wave and second, in showing evidence of frequency processing in the auditory cortex with a machine learning based classification of the EEG signals associated with different frequencies and auditory stimulation modalities. In the control group of twelve healthy volunteers, cortical frequency processing was clearly demonstrated. EEG recordings from two patients with disorders of consciousness showed evidence of partially preserved cortical processing in the first patient and none in the second patient. From these results, it appears that the classification method presented here reliably detects signal differences in the encoding of frequencies and is a useful tool in the evaluation of the integrity of the auditory cortex. Even though the classification method presented in this work was designed for patients with disorders of consciousness, it can also be applied to other pathological populations.