Gert Cauwenberghs

Neuromorphic silicon neuron circuits

Hardware implementations of spiking neurons can be extremely useful for a large variety of applications, ranging from high-speed modeling of large-scale neural systems to real-time behaving systems, to bidirectional brain–machine interfaces. The specific circuit solutions used to implement silicon neurons depend on the application requirements. In this paper we describe the most common building blocks and techniques used to implement these circuits, and present an overview of a wide range of neuromorphic silicon neurons, which implement different computational models, ranging from biophysically realistic and conductance-based Hodgkin–Huxley models to bi-dimensional generalized adaptive integrate and fire models. We compare the different design methodologies used for each silicon neuron design described, and demonstrate their features with experimental results, measured from a wide range of fabricated VLSI chips.

Dry-Contact and Noncontact Biopotential Electrodes: Methodological Review

Recent demand and interest in wireless, mobile-based healthcare has driven significant interest towards developing alternative biopotential electrodes for patient physiological monitoring. The conventional wet adhesive Ag/AgCl electrodes used almost universally in clinical applications today provide an excellent signal but are cumbersome and irritating for mobile use. While electrodes that operate without gels, adhesives and even skin contact have been known for many decades, they have yet to achieve any acceptance for medical use. In addition, detailed knowledge and comparisons between different electrodes are not well known in the literature. In this paper, we explore the use of dry/noncontact electrodes for clinical use by first explaining the electrical models for dry, insulated and noncontact electrodes and show the performance limits, along with measured data. The theory and data show that the common practice of minimizing electrode resistance may not always be necessary and actually lead to increased noise depending on coupling capacitance. Theoretical analysis is followed by an extensive review of the latest dry electrode developments in the literature. The paper concludes with highlighting some of the novel systems that dry electrode technology has enabled for cardiac and neural monitoring followed by a discussion of the current challenges and a roadmap going forward.

Dynamically Reconfigurable Silicon Array of Spiking Neurons With Conductance-Based Synapses

A mixed-signal very large scale integration (VLSI) chip for large scale emulation of spiking neural networks is presented. The chip contains 2400 silicon neurons with fully programmable and reconfigurable synaptic connectivity. Each neuron implements a discrete-time model of a single-compartment cell. The model allows for analog membrane dynamics and an arbitrary number of synaptic connections, each with tunable conductance and reversal potential. The array of silicon neurons functions as an address-event (AE) transceiver, with incoming and outgoing spikes communicated over an asynchronous event-driven digital bus. Address encoding and conflict resolution of spiking events are implemented via a randomized arbitration scheme that ensures balanced servicing of event requests across the array. Routing of events is implemented externally using dynamically programmable random-access memory that stores a postsynaptic address, the conductance, and the reversal potential of each synaptic connection. Here, we describe the silicon neuron circuits, present experimental data characterizing the 3 mm times 3 mm chip fabricated in 0.5-mum complementary metal-oxide-semiconductor (CMOS) technology, and demonstrate its utility by configuring the hardware to emulate a model of attractor dynamics and waves of neural activity during sleep in rat hippocampus

SVM incremental learning, adaptation and optimization

The objective of machine learning is to identify a model that yields good generalization performance. This involves repeatedly selecting a hypothesis class, searching the hypothesis class by minimizing a given objective function over the models parameter space, and evaluating the generalization performance of the resulting model. This search can be computationally intensive as training data continuously arrives, or as one needs to tune hyperparameters in the hypothesis class and the objective function. In this paper, we present a framework for exact incremental learning and adaptation of support vector machine (SVM) classifiers. The approach is general and allows one to learn and unlearn individual or multiple examples, adapt the current SVM to changes in regularization and kernel parameters, and evaluate generalization performance through exact leave-one-out error estimation.

Micropower CMOS Integrated Low-Noise Amplification, Filtering, and Digitization of Multimodal Neuropotentials

Electrical activity in the brain spans a wide range of spatial and temporal scales, requiring simultaneous recording of multiple modalities of neurophysiological signals in order to capture various aspects of brain state dynamics. Here, we present a 16-channel neural interface integrated circuit fabricated in a 0.5 mum 3M2P CMOS process for selective digital acquisition of biopotentials across the spectrum of neural signal modalities in the brain, ranging from single spike action potentials to local field potentials (LFP), electrocorticograms (ECoG), and electroencephalograms (EEG). Each channel is composed of a tunable bandwidth, fixed gain front-end amplifier and a programmable gain/resolution continuous-time incremental DeltaSigma analog-to-digital converter (ADC). A two-stage topology for the front-end voltage amplifier with capacitive feedback offers independent tuning of the amplifier bandpass frequency corners, and attains a noise efficiency factor (NEF) of 2.9 at 8.2 kHz bandwidth for spike recording, and a NEF of 3.2 at 140 Hz bandwidth for EEG recording. The amplifier has a measured midband gain of 39.6 dB, frequency response from 0.2 Hz to 8.2 kHz, and an input-referred noise of 1.94 muV rms while drawing 12.2 muA of current from a 3.3 V supply. The lower and higher cutoff frequencies of the bandpass filter are adjustable from 0.2 to 94 Hz and 140 Hz to 8.2 kHz, respectively. At 10-bit resolution, the ADC has an SNDR of 56 dB while consuming 76 muW power. Time-modulation feedback in the ADC offers programmable digital gain (1-4096) for auto-ranging, further improving the dynamic range and linearity of the ADC. Experimental recordings with the system show spike signals in rat somatosensory cortex as well as alpha EEG activity in a human subject.

Dry and Noncontact EEG Sensors for Mobile Brain–Computer Interfaces

Dry and noncontact electroencephalographic (EEG) electrodes, which do not require gel or even direct scalp coupling, have been considered as an enabler of practical, real-world, brain-computer interface (BCI) platforms. This study compares wet electrodes to dry and through hair, noncontact electrodes within a steady state visual evoked potential (SSVEP) BCI paradigm. The construction of a dry contact electrode, featuring fingered contact posts and active buffering circuitry is presented. Additionally, the development of a new, noncontact, capacitive electrode that utilizes a custom integrated, high-impedance analog front-end is introduced. Offline tests on 10 subjects characterize the signal quality from the different electrodes and demonstrate that acquisition of small amplitude, SSVEP signals is possible, even through hair using the new integrated noncontact sensor. Online BCI experiments demonstrate that the information transfer rate (ITR) with the dry electrodes is comparable to that of wet electrodes, completely without the need for gel or other conductive media. In addition, data from the noncontact electrode, operating on the top of hair, show a maximum ITR in excess of 19 bits/min at 100% accuracy (versus 29.2 bits/min for wet electrodes and 34.4 bits/min for dry electrodes), a level that has never been demonstrated before. The results of these experiments show that both dry and noncontact electrodes, with further development, may become a viable tool for both future mobile BCI and general EEG applications.

Kerneltron: support vector "machine" in silicon

Detection of complex objects in streaming video poses two fundamental challenges: training from sparse data with proper generalization across variations in the object class and the environment; and the computational power required of the trained classifier running real-time. The Kerneltron supports the generalization performance of a support vector machine (SVM) and offers the bandwidth and efficiency of a massively parallel architecture. The mixed-signal very large-scale integration (VLSI) processor is dedicated to the most intensive of SVM operations: evaluating a kernel over large numbers of vectors in high dimensions. At the core of the Kerneltron is an internally analog, fine-grain computational array performing externally digital inner-products between an incoming vector and each of the stored support vectors. The three-transistor unit cell in the array combines single-bit dynamic storage, binary multiplication, and zero-latency analog accumulation. Precise digital outputs are obtained through oversampled quantization of the analog array outputs combined with bit-serial unary encoding of the digital inputs. The 256 input, 128 vector Kerneltron measures 3 mm/spl times/3mm in 0.5 /spl mu/m CMOS, delivers 6.5 GMACS throughput at 5.9 mW power, and attains 8-bit output resolution.

VLSI Potentiostat Array With Oversampling Gain Modulation for Wide-Range Neurotransmitter Sensing

A 16-channel current-measuring very large-scale integration (VLSI) sensor array system for highly sensitive electrochemical detection of electroactive neurotransmiters like dopamine and nitric-oxide is presented. Each channel embeds a current integrating potentiostat within a switched-capacitor first-order single-bit delta-sigma modulator implementing an incremental analog-to-digital converter. The duty-cycle modulation of current feedback in the delta-sigma loop together with variable oversampling ratio provide a programmable digital range selection of the input current spanning over six orders of magnitude from picoamperes to microamperes. The array offers 100-fA input current sensitivity at 3.4-muW power consumption per channel. The operation of the 3 mm times3 mm chip fabricated in 0.5-mum CMOS technology is demonstrated with real-time multichannel acquisition of neurotransmitter concentration

An analog VLSI recurrent neural network learning a continuous-time trajectory

Real-time algorithms for gradient descent supervised learning in recurrent dynamical neural networks fail to support scalable VLSI implementation, due to their complexity which grows sharply with the network dimension. We present an alternative implementation in analog VLSI, which employs a stochastic perturbation algorithm to observe the gradient of the error index directly on the network in random directions of the parameter space, thereby avoiding the tedious task of deriving the gradient from an explicit model of the network dynamics. The network contains six fully recurrent neurons with continuous-time dynamics, providing 42 free parameters which comprise connection strengths and thresholds. The chip implementing the network includes local provisions supporting both the learning and storage of the parameters, integrated in a scalable architecture which can be readily expanded for applications of learning recurrent dynamical networks requiring larger dimensionality. We describe and characterize the functional elements comprising the implemented recurrent network and integrated learning system, and include experimental results obtained from training the network to represent a quadrature-phase oscillator.

A Low-Noise, Non-Contact EEG/ECG Sensor

Typical electroencephalogram (EEG) and electrocardiogram (ECG) sensors require conductive gel to ensure low-impedance electrical contact between the sensor and skin, making set-up time-consuming and long-term recording problematic. We present a gel-free, non-contact EEG/ECG sensor with on-board electrode that capacitively couples to the skin. Active shielding of the high-impedance input significantly reduces noise pickup, and reduces variations in gain as a function of gap distance. The integrated sensor combines amplification, bandpass filtering, and analog-to-digital conversion within a 1 inch diameter enclosure. The measured input-referred noise, over 1-100 Hz frequency range, is 2 muVrms at 0.2 mm sensor distance, and 17 muVrms at 3.2 mm distance. Experiments coupling the sensor to human scalp through hair and to chest through clothing produce clear EEG and ECG recorded signals.