Searcharxiv⌕ Search

arXiv subjects

Eric Müller

Publications and source records attributed to Eric Müller.

At least 37 records · Page 2Linked to original sources
arXiv2022

hxtorch.snn: Machine-learning-inspired Spiking Neural Network Modeling on BrainScaleS-2

Neuromorphic systems require user-friendly software to support the design and optimization of experiments. In this work, we address this need by presenting our development of a machine learning-based modeling framework for the BrainScaleS-2 neuromorphic system. This work represents an improvement over previous efforts, which either focused on the matrix-multiplication mode of BrainScaleS-2 or lacked full automation. Our framework, called hxtorch.snn, enables the hardware-in-the-loop training of spiking neural networks within PyTorch, including support for auto differentiation in a fully-automated hardware experiment workflow. In addition, hxtorch.snn facilitates seamless transitions between emulating on hardware and simulating in software. We demonstrate the capabilities of hxtorch.snn on a classification task using the Yin-Yang dataset employing a gradient-based approach with surrogate gradients and densely sampled membrane observations from the BrainScaleS-2 hardware system.

cs.NE↗
arXiv2022

Demonstrating Analog Inference on the BrainScaleS-2 Mobile System

We present the BrainScaleS-2 mobile system as a compact analog inference engine based on the BrainScaleS-2 ASIC and demonstrate its capabilities at classifying a medical electrocardiogram dataset. The analog network core of the ASIC is utilized to perform the multiply-accumulate operations of a convolutional deep neural network. At a system power consumption of 5.6W, we measure a total energy consumption of 192uJ for the ASIC and achieve a classification time of 276us per electrocardiographic patient sample. Patients with atrial fibrillation are correctly identified with a detection rate of (93.7${\pm}$0.7)% at (14.0${\pm}$1.0)% false positives. The system is directly applicable to edge inference applications due to its small size, power envelope, and flexible I/O capabilities. It has enabled the BrainScaleS-2 ASIC to be operated reliably outside a specialized lab setting. In future applications, the system allows for a combination of conventional machine learning layers with online learning in spiking neural networks on a single neuromorphic platform.

cs.AR↗
arXiv2022

Autocorrelations from emergent bistability in homeostatic spiking neural networks on neuromorphic hardware

A unique feature of neuromorphic computing is that memory is an implicit part of processing through traces of past information in the system's collective dynamics. The extent of memory about past inputs is commonly quantified by the autocorrelation time of collective dynamics. Based on past experimental evidence, a potential explanation for the underlying autocorrelations are close-to-critical fluctuations. Here, we show for self-organized networks of excitatory and inhibitory leaky integrate-and-fire neurons that autocorrelations can originate from emergent bistability upon reducing external input strength. We identify the bistability as a fluctuation-induced stochastic switching between metastable active and quiescent states in the vicinity of a non-equilibrium phase transition. This bistability occurs for networks with fixed heterogeneous weights as a consequence of homeostatic self-organization during development. Specifically, in our experiments on neuromorphic hardware and in computer simulations, the emergent bistability gives rise to autocorrelation times exceeding 500 ms despite single-neuron timescales of only 20 ms. Our results provide the first verification of biologically compatible autocorrelation times in networks of leaky integrate-and-fire neurons, which here are not generated by close-to-critical fluctuations but by emergent bistability in homeostatically regulated networks. Our results thereby constitute a new, complementary mechanism for emergent autocorrelations in networks of spiking neurons, with implications for biological and artificial networks, and introduces the general paradigm of fluctuation-induced bistability for driven systems with absorbing states.

q-bio.NC↗
arXiv2022

Versatile emulation of spiking neural networks on an accelerated neuromorphic substrate

We present first experimental results on the novel BrainScaleS-2 neuromorphic architecture based on an analog neuro-synaptic core and augmented by embedded microprocessors for complex plasticity and experiment control. The high acceleration factor of 1000 compared to biological dynamics enables the execution of computationally expensive tasks, by allowing the fast emulation of long-duration experiments or rapid iteration over many consecutive trials. The flexibility of our architecture is demonstrated in a suite of five distinct experiments, which emphasize different aspects of the BrainScaleS-2 system.

q-bio.NC↗
arXiv2022

A Scalable Approach to Modeling on Accelerated Neuromorphic Hardware

Neuromorphic systems open up opportunities to enlarge the explorative space for computational research. However, it is often challenging to unite efficiency and usability. This work presents the software aspects of this endeavor for the BrainScaleS-2 system, a hybrid accelerated neuromorphic hardware architecture based on physical modeling. We introduce key aspects of the BrainScaleS-2 Operating System: experiment workflow, API layering, software design, and platform operation. We present use cases to discuss and derive requirements for the software and showcase the implementation. The focus lies on novel system and software features such as multi-compartmental neurons, fast re-configuration for hardware-in-the-loop training, applications for the embedded processors, the non-spiking operation mode, interactive platform access, and sustainable hardware/software co-development. Finally, we discuss further developments in terms of hardware scale-up, system usability and efficiency.

cs.NE↗
arXiv2022

Demonstrating BrainScaleS-2 Inter-Chip Pulse-Communication using EXTOLL

The BrainScaleS-2 (BSS-2) Neuromorphic Computing System currently consists of multiple single-chip setups, which are connected to a compute cluster via Gigabit-Ethernet network technology. This is convenient for small experiments, where the neural networks fit into a single chip. When modeling networks of larger size, neurons have to be connected across chip boundaries. We implement these connections for BSS-2 using the EXTOLL networking technology. This provides high bandwidths and low latencies, as well as high message rates. Here, we describe the targeted pulse-routing implementation and required extensions to the BSS-2 software stack. We as well demonstrate feed-forward pulse-routing on BSS-2 using a scaled-down version without temporal merging.

cs.AR↗
arXiv2022

The BrainScaleS-2 accelerated neuromorphic system with hybrid plasticity

Since the beginning of information processing by electronic components, the nervous system has served as a metaphor for the organization of computational primitives. Brain-inspired computing today encompasses a class of approaches ranging from using novel nano-devices for computation to research into large-scale neuromorphic architectures, such as TrueNorth, SpiNNaker, BrainScaleS, Tianjic, and Loihi. While implementation details differ, spiking neural networks - sometimes referred to as the third generation of neural networks - are the common abstraction used to model computation with such systems. Here we describe the second generation of the BrainScaleS neuromorphic architecture, emphasizing applications enabled by this architecture. It combines a custom analog accelerator core supporting the accelerated physical emulation of bio-inspired spiking neural network primitives with a tightly coupled digital processor and a digital event-routing network.

cs.NE↗
arXiv2022

The Operating System of the Neuromorphic BrainScaleS-1 System

BrainScaleS-1 is a wafer-scale mixed-signal accelerated neuromorphic system targeted for research in the fields of computational neuroscience and beyond-von-Neumann computing. The BrainScaleS Operating System (BrainScaleS OS) is a software stack giving users the possibility to emulate networks described in the high-level network description language PyNN with minimal knowledge of the system. At the same time, expert usage is facilitated by allowing to hook into the system at any depth of the stack. We present operation and development methodologies implemented for the BrainScaleS-1 neuromorphic architecture and walk through the individual components of BrainScaleS OS constituting the software stack for BrainScaleS-1 platform operation.

cs.NE↗
arXiv2021

BrainScaleS Large Scale Spike Communication using Extoll

The BrainScaleS Neuromorphic Computing System is currently connected to a compute cluster via Gigabit-Ethernet network technology. This is convenient for the currently used experiment mode, where neuronal networks cover at most one wafer module. When modelling networks of larger size, as for example a full sized cortical microcircuit model, one has to think about connecting neurons across wafer modules to larger networks. This can be done, using the Extoll networking technology, which provides high bandwidth and low latencies, as well as a low overhead packet protocol format.

cs.AR↗
arXiv2020

Detector Systems Engineering for Extremely Large Instruments

The scientific detector systems for the ESO ELT first-light instruments, HARMONI, MICADO, and METIS, together will require 27 science detectors: seventeen 2.5 $μ$m cutoff H4RG-15 detectors, four 4K x 4K 231-84 CCDs, five 5.3 $μ$m cutoff H2RG detectors, and one 13.5 $μ$m cutoff GEOSNAP detector. This challenging program of scientific detector system development covers everything from designing and producing state-of-the-art detector control and readout electronics, to developing new detector characterization techniques in the lab, to performance modeling and final system verification. We report briefly on the current design of these detector systems and developments underway to meet the challenging scientific performance goals of the ELT instruments.

astro-ph.IM↗
arXiv2020

Infrared wavefront sensing for adaptive optics assisted Galactic Center observations with the VLT interferometer and GRAVITY: operation and results

This article describes the operation of the near-infrared wavefront sensing based Adaptive Optics (AO) system CIAO. The Coudé Infrared Adaptive Optics (CIAO) system is a central auxiliary component of the Very Large Telescope (VLT) interferometer (VLTI). It enables in particular the observations of the Galactic Center (GC) using the GRAVITY instrument. GRAVITY is a highly specialized beam combiner, a device that coherently combines the light of the four 8-m telescopes and finally records interferometric measurements in the K-band on 6 baselines simultaneously. CIAO compensates for phase disturbances caused by atmospheric turbulence, which all four 8 m Unit Telescopes (UT) experience during observation. Each of the four CIAO units generates an almost diffraction-limited image quality at its UT, which ensures that maximum flux of the observed stellar object enters the fibers of the GRAVITY beam combiner. We present CIAO performance data obtained in the first 3 years of operation as a function of weather conditions. We describe how CIAO is configured and used for observations with GRAVITY. In addition, we focus on the outstanding features of the near-infrared sensitive Saphira detector, which is used for the first time on Paranal, and show how it works as a wavefront sensor detector.

astro-ph.IM↗
arXiv2020

hxtorch: PyTorch for BrainScaleS-2 -- Perceptrons on Analog Neuromorphic Hardware

We present software facilitating the usage of the BrainScaleS-2 analog neuromorphic hardware system as an inference accelerator for artificial neural networks. The accelerator hardware is transparently integrated into the PyTorch machine learning framework using its extension interface. In particular, we provide accelerator support for vector-matrix multiplications and convolutions; corresponding software-based autograd functionality is provided for hardware-in-the-loop training. Automatic partitioning of neural networks onto one or multiple accelerator chips is supported. We analyze implementation runtime overhead during training as well as inference, provide measurements for existing setups and evaluate the results in terms of the accelerator hardware design limitations. As an application of the introduced framework, we present a model that classifies activities of daily living with smartphone sensor data.

cs.NE↗
arXiv2020

Inference with Artificial Neural Networks on Analog Neuromorphic Hardware

The neuromorphic BrainScaleS-2 ASIC comprises mixed-signal neurons and synapse circuits as well as two versatile digital microprocessors. Primarily designed to emulate spiking neural networks, the system can also operate in a vector-matrix multiplication and accumulation mode for artificial neural networks. Analog multiplication is carried out in the synapse circuits, while the results are accumulated on the neurons' membrane capacitors. Designed as an analog, in-memory computing device, it promises high energy efficiency. Fixed-pattern noise and trial-to-trial variations, however, require the implemented networks to cope with a certain level of perturbations. Further limitations are imposed by the digital resolution of the input values (5 bit), matrix weights (6 bit) and resulting neuron activations (8 bit). In this paper, we discuss BrainScaleS-2 as an analog inference accelerator and present calibration as well as optimization strategies, highlighting the advantages of training with hardware in the loop. Among other benchmarks, we classify the MNIST handwritten digits dataset using a two-dimensional convolution and two dense layers. We reach 98.0% test accuracy, closely matching the performance of the same network evaluated in software.

cs.NE↗
arXiv2020

Accelerated physical emulation of Bayesian inference in spiking neural networks

The massively parallel nature of biological information processing plays an important role for its superiority to human-engineered computing devices. In particular, it may hold the key to overcoming the von Neumann bottleneck that limits contemporary computer architectures. Physical-model neuromorphic devices seek to replicate not only this inherent parallelism, but also aspects of its microscopic dynamics in analog circuits emulating neurons and synapses. However, these machines require network models that are not only adept at solving particular tasks, but that can also cope with the inherent imperfections of analog substrates. We present a spiking network model that performs Bayesian inference through sampling on the BrainScaleS neuromorphic platform, where we use it for generative and discriminative computations on visual data. By illustrating its functionality on this platform, we implicitly demonstrate its robustness to various substrate-specific distortive effects, as well as its accelerated capability for computation. These results showcase the advantages of brain-inspired physical computation and provide important building blocks for large-scale neuromorphic applications.

cs.NE↗
arXiv2020

Extending BrainScaleS OS for BrainScaleS-2

BrainScaleS-2 is a mixed-signal accelerated neuromorphic system targeted for research in the fields of computational neuroscience and beyond-von-Neumann computing. To augment its flexibility, the analog neural network core is accompanied by an embedded SIMD microprocessor. The BrainScaleS Operating System (BrainScaleS OS) is a software stack designed for the user-friendly operation of the BrainScaleS architectures. We present and walk through the software-architectural enhancements that were introduced for the BrainScaleS-2 architecture. Finally, using a second-version BrainScaleS-2 prototype we demonstrate its application in an example experiment based on spike-based expectation maximization.

cs.NE↗
arXiv2020

Nightside condensation of iron in an ultra-hot giant exoplanet

Ultra-hot giant exoplanets receive thousands of times Earth's insolation. Their high-temperature atmospheres (>2,000 K) are ideal laboratories for studying extreme planetary climates and chemistry. Daysides are predicted to be cloud-free, dominated by atomic species and substantially hotter than nightsides. Atoms are expected to recombine into molecules over the nightside, resulting in different day-night chemistry. While metallic elements and a large temperature contrast have been observed, no chemical gradient has been measured across the surface of such an exoplanet. Different atmospheric chemistry between the day-to-night ("evening") and night-to-day ("morning") terminators could, however, be revealed as an asymmetric absorption signature during transit. Here, we report the detection of an asymmetric atmospheric signature in the ultra-hot exoplanet WASP-76b. We spectrally and temporally resolve this signature thanks to the combination of high-dispersion spectroscopy with a large photon-collecting area. The absorption signal, attributed to neutral iron, is blueshifted by -11+/-0.7 km s-1 on the trailing limb, which can be explained by a combination of planetary rotation and wind blowing from the hot dayside. In contrast, no signal arises from the nightside close to the morning terminator, showing that atomic iron is not absorbing starlight there. Iron must thus condense during its journey across the nightside.

astro-ph.EP↗