J.M. Rabaey

• PhD
• Professor at University of California, Berkeley

The remarkable advancements in artificial intelligence (AI), primarily driven by deep neural networks, are facing challenges surrounding unsustainable computational trajectories, limited robustness, and a lack of explainability. To develop next-generation cognitive AI systems, neuro-symbolic AI emerges as a promising paradigm, fusing neural and sym...

The remarkable advancements in artificial intelligence (AI), primarily driven by deep neural networks, are facing challenges surrounding unsustainable computational trajectories, limited robustness, and a lack of explainability. To develop next-generation cognitive AI systems, neuro-symbolic AI emerges as a promising paradigm, fusing neural and sym...

In this paper, a hardware-optimized approach to emotion recognition based on the efficient brain-inspired hyperdimensional computing (HDC) paradigm is proposed. Emotion recognition provides valuable information for human–computer interactions; however, the large number of input channels (> 200) and modalities (> 3 ) involved in emotion recognition...

This article reviews recent progress in the development of the computing framework vector symbolic architectures (VSA) (also known as hyperdimensional computing). This framework is well suited for implementation in stochastic, emerging hardware, and it naturally expresses the types of cognitive operations required for artificial intelligence (AI)....

We introduce a method to identify speakers by computing with high-dimensional random vectors. Its strengths are simplicity and speed. With only 1.02k active parameters and a 128-minute pass through the training data we achieve Top-1 and Top-5 scores of 31% and 52% on the VoxCeleb1 dataset of 1,251 speakers. This is in contrast to CNN models requiri...

Hyperdimensional computing (HDC) is a brain-inspired computing paradigm that operates on pseudo-random hypervectors to perform high-accuracy classifications for biomedical applications. The energy efficiency of prior HDC processors for this computationally minimal algorithm is dominated by costly hypervector memory storage, which grows linearly wit...

Surface-mounted parallel-plate capacitors (PPCs) can be used as transducing elements in wireless communication, for example, in human-body-dedicated applications, such as healthcare monitoring or activity tracking, which require robust wireless communications and user-convenient implementation. In this context, PPCs are an attractive element due to...

Traditional methods for radio frequency (RF) beamforming require significant amounts of power, making them difficult to deploy in some low-power applications. A detailed analysis of RF beamforming utilizing transmission-line transformers (TLTs) and balanced impedance phase shifters (BIPSs) is presented. This technique enables ultralow-power RF beam...

Memory-augmented neural networks enhance a neural network with an external key-value (KV) memory whose complexity is typically dominated by the number of support vectors in the key memory. We propose a generalized KV memory that decouples its dimension from the number of support vectors by introducing a free parameter that can arbitrarily add or re...

Memory-augmented neural networks enhance a neural network with an external key-value memory whose complexity is typically dominated by the number of support vectors in the key memory. We propose a generalized key-value memory that decouples its dimension from the number of support vectors by introducing a free parameter that can arbitrarily add or...

Semiconductor innovation drives improvements to technologies that are critical to modern society. The country that successfully accelerates semiconductor innovation is positioned to lead future semiconductor-driven industries and benefit from the resulting economic growth. It is our view that a next generation infrastructure is necessary to acceler...

div>Prosthetic control for rehabilitation, among many other applications, can leverage in-sensor hand gesture recognition in which lightweight machine learning models for classifying electromyogram (EMG) signals are embedded on miniature, low-power devices. While research efforts have demonstrated high accuracy in controlled settings, these methods...

div>Prosthetic control for rehabilitation, among many other applications, can leverage in-sensor hand gesture recognition in which lightweight machine learning models for classifying electromyogram (EMG) signals are embedded on miniature, low-power devices. While research efforts have demonstrated high accuracy in controlled settings, these methods...

Machine learning algorithms deployed on edge devices must meet certain resource constraints and efficiency requirements. Random Vector Functional Link (RVFL) networks are favored for such applications due to their simple design and training efficiency. We propose a modified RVFL network that avoids computationally expensive matrix operations during...

Machine learning algorithms deployed on edge devices must meet certain resource constraints and efficiency requirements. Random Vector Functional Link (RVFL) networks are favored for such applications due to their simple design and training efficiency. We propose a modified RVFL network that avoids computationally expensive matrix operations during...

This article reviews recent progress in the development of the computing framework Vector Symbolic Architectures (also known as Hyperdimensional Computing). This framework is well suited for implementation in stochastic, nanoscale hardware and it naturally expresses the types of cognitive operations required for Artificial Intelligence (AI). We dem...

In this paper, a hardware-optimized approach to emotion recognition based on the efficient brain-inspired hyperdimensional computing (HDC) paradigm is proposed. Emotion recognition provides valuable information for human-computer interactions, however the large number of input channels (>200) and modalities (>3) involved in emotion recognition are...

Wireless body area networks (WBANs) present unique challenges due to their specific characteristics of mobility and over-the-body radio propagation. A huge amount of factors—both internal and external to the network—affect WBAN channel conditions, so a reliable and comprehensive theoretical model of these communications is unfeasible and impractica...

Electromyogram (EMG) pattern recognition can be used to classify hand gestures and movements for human-machine interface and prosthetics applications, but it often faces reliability issues resulting from limb position change. One method to address this is dual-stage classification, in which the limb position is first determined using additional sen...

Wearable devices that monitor muscle activity based on surface electromyography could be of use in the development of hand gesture recognition applications. Such devices typically use machine-learning models, either locally or externally, for gesture classification. However, most devices with local processing cannot offer training and updating of t...

Sensor data can be wirelessly transmitted from simple, battery-less tags using Radio Frequency Identification (RFID). RFID sensor tags consist of an antenna, a radio frequency integrated circuit chip (RFIC), and at least one sensor. An ideal tag can communicate over a long distance and be seamlessly integrated onto everyday objects. However, miniat...

Wireless Body Area Networks (WBANs) are a fast-growing field fueled by the number of wearable devices developed for countless applications appearing on the market. To enable communication between a variety of those devices, the IEEE 802.15.6 standard was established. However, this standard has some intrinsic limitations in addressing the heterogene...

The increasing penetration of wearable and implantable devices necessitates energy-efficient and robust ways of connecting them to each other and to the cloud. However, the wireless channel around the human body poses unique challenges such as a high and variable path-loss caused by frequent changes in the relative node positions as well as the sur...

With the world around us rapidly becoming smarter, an extremely relevant question is how “we humans” are going to cope with the onslaught of information coming at us. One plausible answer is to use similar technologies to evolve ourselves and to equip us with the necessary tools to interact with and to become an essential part of the smart world. V...

One viable solution for continuous reduction in energy-per-operation is to rethink functionality to cope with uncertainty by adopting computational approaches that are inherently robust to uncertainty. It requires a novel look at data representations, associated operations, and circuits, and at materials and substrates that enable them. 3D integrat...

Closed-loop neuromodulation systems aim to treat a variety of neurological conditions by delivering and adjusting therapeutic electrical stimulation in response to a patient’s neural state, recorded in real time. Existing systems are limited by low channel counts, lack of algorithmic flexibility, and the distortion of recorded signals by large and...

Accurate recognition of hand gestures is crucial to the functionality of smart prosthetics and other modern human-computer interfaces. Many machine learning-based classifiers use electromyography (EMG) signals as input features, but they often misclassify gestures performed in different situational contexts (changing arm position, reapplication of...

One viable solution for continuous reduction in energy-per-operation is to rethink functionality to cope with uncertainty by adopting computational approaches that are inherently robust to uncertainty. It requires a novel look at data representations, associated operations, and circuits, and at materials and substrates that enable them. 3D integrat...

This paper investigates the potential of using spray coating as a methodology for flexible antenna fabrication. The methodology has advantages compared to other antenna-printing techniques such as screen-printing and gravure printing (more flexibility in design), or inkjet printing (faster production). The methodology is demonstrated using two diff...

Recognizing the very size of the brain’s circuits, hyperdimensional (HD) computing can model neural activity patterns with points in a HD space, that is, with HD vectors. Key examined properties of HD computing include: a versatile set of arithmetic operations on HD vectors, generality, scalability, analyzability, one-shot learning, and energy effi...

The field of machine learning is witnessing rapid advances along several fronts: new machine learning models, new machine learning algorithms utilizing these models, new hardware architectures for these algorithms, and new technologies for creating energy-efficient implementations of such hardware architectures. Hyperdimensional (HD) computing repr...

We present a 65-nm CMOS neuromodulation SoC with a stimulus artifact canceler allowing simultaneous recording and stimulation. A first-order ΔΣ-modulated switched capacitor DAC injects artifact canceling charges at the LNA input, handling up to 200 mV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">pp<...

The Human Intranet is envisioned as an open, scalable platform that seamlessly integrates an ever-increasing number of sensor, actuation, computation, storage, communication, and energy nodes located on, in, or around the human body, acting in symbiosis with the functions provided by the body itself. The limited amount of available energy and the c...

An ultra-wideband (UWB)-based cognitive radio (CR) is a promising technique to utilize 3.1-10.6-GHz band efficiently for high data-rate short-range wireless connectivity even in overcrowdedness of the frequency spectrum. Frequency synthesizers tailored to the UWB-based CR should provide wide lock range, multiple frequency bands, and low in-band spu...

The emergence of post-silicon nano-devices and the coming trillion-sensor era driven by the Internet of Things (IoT) have led to a search for alternative computational paradigms that can efficiently derive useful information from abundant data while enable efficient hardware implementations under significant device variations. Such computational pa...

Hyperdimensional (HD) computing is a promising paradigm for future intelligent electronic appliances operating at low power. This paper discusses tradeoffs of selecting parameters of binary HD representations when applied to pattern recognition tasks. Particular design choices include density of representations and strategies for mapping data from...

EMG-based gesture recognition shows promise for human-machine interaction. Systems are often afflicted by signal and electrode variability which degrades performance over time. We present an end-to-end system combating this variability using a large-area, high-density sensor array and a robust classification algorithm. EMG electrodes are fabricated...

An ultra-wideband (UWB)-based cognitive radio (CR) is a promising technique to utilize 3.1-10.6 GHz band efficiently for high data-rate short-range wireless connectivity even in the overcrowded spectrum. This paper investigates a low power wavelet-based energy detection spectrum sensor that provides high-order band-pass filter function and spectral...

Location information of mobile devices is a foun-dational input to location-based services and a valuable source of context information in wireless networks. To maximize the value, we need location information that is accurate, robust, and promptly and seamlessly available. Unfortunately, individual localization services seldom satisfy all these re...

Closed-loop neuromodulation systems aim to treat a variety of neurological conditions by dynamically delivering and adjusting therapeutic electrical stimulation in response to the neural state of a patient. Closed-loop devices must simultaneously record neural signals, remove stimulation artifact from recorded data, and extract neural biomarkers to...

We address the design space exploration of wireless body area networks for wearable and implantable technologies, a task that is increasingly challenging as the number and variety of devices per person grow. Our method efficiently decomposes the problem into smaller subproblems by coordinating specialized analysis and optimization techniques. We le...

This review focuses on recent directions stemming from work by the authors and collaborators in the emerging field of neurotechnology. Neurotechnology has the potential to provide a greater understanding of the structure and function of the complex neural circuits in the brain, as well as impacting the field of brain–machine interfaces (BMI). We en...

Localization services available on today's mobile devices are proprietary and leverage a limited set of sources of location information. Integration of new location estimation methods is therefore cumbersome, requiring adaptation to the specific interfaces of the proprietary location service. In addition , location-based applications are tightly in...

The emerging field of bioelectronic medicine seeks methods for deciphering and modulating electrophysiological activity in the body to attain therapeutic effects at target organs. Current approaches to interfacing with peripheral nerves and muscles rely heavily on wires, creating problems for chronic use, while emerging wireless approaches lack the...

A distributed, modular, intelligent, and efficient neuromodulation device, called OMNI, is presented. It supports closed-loop recording and stimulation on 256 channels from up to 4 physically distinct neuromodulation modules placed in any configuration around the brain, hence offering the capability of addressing neural disorders that are presented...

In this paper, we present an ultrasonic beamforming system capable of interrogating individual implantable sensors via backscatter in a distributed, ultrasound-based recording platform known as Neural Dust [1]. A custom ASIC drives a 7 × 2 PZT transducer array with 3 cycles of 32V square wave with a specific programmable time delay to focus the bea...

A triple-channel BPSK UWB-based cognitive radio provides energy efficient 1Gb/s short-range connectivity by scavenging triple discrete inactive frequency bands in 3.1–10.6GHz ISM bands. The developed transceiver in 65nm CMOS achieves the minimum energy consumption of 59.7pJ/b with 1.97×10 −4 BER. Die area is 4.6mm 2 with on-die PLLs.

A 65 nm CMOS 4.78 mm 2 integrated neuromodulation SoC consumes 348 µA from an unregulated 1.2 V to 1.8 V supply while operating 64 acquisition channels with epoch compression at an average firing rate of 50 Hz and engaging two stimulators with a pulse width of 250 µs/phase, differential current of 150 µA, and a pulse frequency of 100 Hz. Compared t...

Brain-machine interface (BMI) technology has tremendous potential to revolutionize healthcare by greatly im-proving the quality of life of millions of people suffering from a wide variety of neurological conditions. Radio-frequency identi-fication (RFID)-inspired backscattering is a promising approach for wireless powering of miniature neural senso...

?Pub Dtl=""?> Emerging applications in brain–machine interface systems require high-resolution, chronic multisite cortical recordings, which cannot be obtained with existing technologies due to high power consumption, high invasiveness, or inability to transmit data wirelessly. In this paper, we describe a microsystem based on electrocorticography...

Fritsch and Hitzig first discovered the motor cortex in 1870 [1], although the best-known experimentalmapping of the motor cortex dates back to Penfield’s experiments in 1937 [2] using electrical stimulation to activate muscle groups in patients undergoing surgery for epilepsy. It was not until the 1980s, over 100 years since the discovery of the m...

A major hurdle in brain-machine interfaces (BMI) is the lack of an implantable neural interface system that remains viable for a substantial fraction of the user's lifetime. Recently, sub-mm implantable, wireless electromagnetic (EM) neural interfaces have been demonstrated in an effort to extend system longevity. However, EM systems do not scale d...

In this paper, we examine the use of beamforming techniques to interrogate a multitude of neural implants in a distributed, ultrasound-based intra-cortical recording platform known as Neural Dust [1]. We propose a general framework to analyze system design tradeoffs in the ultrasonic beamformer that extracts neural signals from modulated ultrasound...

In this paper, a new class of pulsed latches is introduced and experimentally assessed in 65-nm CMOS. Its conditional push–pull pulsed latch topology is based on a push–pull final stage driven by two split paths with a conditional pulse generator. Two circuit implementations of the concept are discussed, with their main difference being in the puls...

Wireless body-centric sensing systems have an important role in the fields of biomedicine, personal healthcare, safety, and security. Body-centric radio-frequency identification (RFID) technology provides a wireless and maintenance-free communication link between the human body and the surroundings through wearable and implanted antennas. This enab...

TerraSwarm applications, or swarmlets, are characterized by their ability to dynamically recruit resources such as sensors, communication networks, computation, and information from the cloud; to aggregate and use that information to make or aid decisions; and then to dynamically recruit actuation resources. The TerraSwarm vision cannot be achieved...

Battery-free cortical implants enable data telemetry in wireless brain-machine interface systems (BMI). In this paper, we analyze the power transfer in a wireless link from a wearable on-body transmit antenna to a miniature 2×2×2-mm3 implant antenna. For the first time, we assess the feasibility of using electro-textiles in the realization of the w...

Substantial improvements in neural-implant longevity are needed to transition brain-machine interface (BMI) systems from research labs to clinical practice. While action potential (AP) recording through penetrating electrode arrays offers the highest spatial resolution, it comes at the price of tissue scarring, resulting in signal degradation over...

Simultaneously measuring the activities of all neurons in a mammalian brain at millisecond resolution is a challenge beyond the limits of existing techniques in neuroscience. Entirely new approaches may be required, motivating an analysis of the fundamental physical constraints on the problem. We outline the physical principles governing brain acti...

We analyze the power and voltage transfer in a wireless link from an on-body transmit antenna to 1×1×1 mm3 antenna in a cortical implant to provide power and data telemetry for a battery-free brain-machine interface microelectronic system. We compare the wireless link performance with regular, segmented, and tilted transmit loop antennas. Moreover,...

A wide lock-range supply regulated integer-N QPLL is proposed to reduce power consumption of the wideband direct digital RF modulator. SINC roll-off characteristic for supply noise of the inverter-based ring-VCOs in frequency domain maximizes loop bandwidth of the wide lock-range PLL. The proposed charge pump keeps loop bandwidth for all integer-N...

Moore's law, the driving force behind the semiconductor for the past decades, is endangered from several angles. Artifacts of scaled dimensions, yield, reliability, fabrication cost and device performance degradation all raise legitimate concerns about current trends of mere transferring same architectures to more advanced substrates.

In this poster abstract, we present recent advances on implantable and wearable antennas for wireless body-centric sensing systems. Prominent examples of such systems are the wireless brain-machine interface (BMI) system and wearable embroidered antennas. In the poster, we present more detailed simulation and measurement results.

A novel reconfigurable switched-capacitor “EChO” Power Management Unit is introduced for ultra-low power duty-cycled integrated systems (e.g., sensor nodes for critical event monitoring). “EChO” reduces the energy cost associated with sleep-to-active and active-to sleep transitions by 64% with an area overhead less than 1% and no impact on active m...

A major hurdle in brain-machine interfaces (BMI) is the lack of an implantable neural interface system that remains viable for a lifetime. This paper explores the fundamental system design trade-offs and ultimate size, power, and bandwidth scaling limits of neural recording systems built from low-power CMOS circuitry coupled with ultrasonic power d...

Backscattering-based RFID-inspired communication provides an effective approach for wireless power and data transfer for implantable battery-free brain machine interface microsystems. We compare conventional and segmented transmit loop configurations to reduce the specific absorption rate (SAR) into human body. Moreover, we present novel cubic-mil...

We present a method for decreasing the duration of artifacts present during intra-cortical microstimulation (ICMS) recordings by using techniques developed for digital communications. We replace the traditional monophasic or biphasic current stimulation pulse with a patterned pulse stream produced by a Zero Forcing Equalizer (ZFE) filter after char...

Although technology scaling of CMOS devices has allowed great advances in device speed and digital computational efficiency, the quality and size of on-chip passive structures have not enjoyed similar scaling. As a result, the power consumption, physical footprint, and performance of modern integrated transceivers are typically limited by their pas...

The brain is an amazingly complex and efficient machine. While it may not be considered "general purpose" in terms of its computational capabilities, it performs a set of functions such as feature extraction, classification, synthesis, recognition, learning, and higher-order decision-making amazingly well. Unfortunately the dynamic behavior of the...

A wirelessly powered 0.125 mm² 65 nm CMOS IC for Brain-Machine Interface applications integrates four 1.5 µW amplifiers (6.5 µVrms input-referred noise with 10 kHz bandwidth) with power conditioning and communication circuitry. The multi-node backscatter frequency locks to a wireless interrogator using a frequency-domain multiple access communicati...

This paper provides a full analysis of powering mm-size cortical implants wirelessly. An effective approach for wireless power and data transfer in neurorecording microsystems is the backscattering-based RFID-inspired communication. In this mechanism, an external interrogator powers the implant unit by microwave radiation power, and the implant IC...

In this letter, we analyze the wireless power transfer from an on-body transmit antenna to a millimeter-size antenna in a cortical implant. The studied wireless link provides power and data telemetry for a battery-free wireless brain-machine interface microelectronic system. We present a cubic 2 × 2 × 2 mm3 implant loop and analyze the effect of a...

Thirty-two years ago, Electronics Magazine honored Carver Mead and Lynn Conway with its Achievement Award for their contributions to VLSI chip design. The ‘Mead & Conway methods’ were being taught at 100+ universities all over the world, and “not only have helped spawn a common design culture so necessary in the VLSI era, but have greatly increased...