Efficient Neuromorphic Signal Processing with Loihi 2

Orchard, Garrick; Frady, E. Paxon; Rubin, Daniel Ben Dayan; Sanborn, Sophia; Shrestha, Sumit Bam; Sommer, Friedrich T.; Davies, Martin

doi:10.48550/arxiv.2111.03746

Cited by 5 publications

(7 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We demonstrate how unsupervised dictionary learning using spiking LCA (S-LCA) can be accomplished by starting with ratecoded non-spiking neurons and employing accumulator neurons to transition the model to a spiking regime. LCA allows for both unsupervised dictionary learning and inference to be performed in a manner compatible with the constraints of recent architectures, such as the Intel Loihi research chip [12].…”

Section: Resultsmentioning

confidence: 99%

“…Some neuromorphic hardware platforms such as SpiNNaker [13] and Loihi 2 [12] offer direct support for discretized spike magnitudes, so elements functionally similar to accumulator neurons may be efficiently supported even with intermediate 𝑠 values where multiple spikes occur per timestep.…”

Section: Accumulator Neuronsmentioning

confidence: 99%

“…Second, there is evidence that biological neural circuits utilize spike timing to transmit information more rapidly and to dynamically bind distributed features via synchronous oscillations [17] [21] [19] [6]. The potential for mimicking the dynamics of biological neural networks in fast, low-power neuromorphic processors has motivated several efforts to develop such devices [12] [5] [1] [13].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Dictionary Learning with Accumulator Neurons

Parpart¹,

González²,

Stewart³

et al. 2022

Preprint

View full text Add to dashboard Cite

The Locally Competitive Algorithm (LCA) uses local competition between non-spiking leaky integrator neurons to infer sparse representations, allowing for potentially real-time execution on massively parallel neuromorphic architectures such as Intel's Loihi processor. Here, we focus on the problem of inferring sparse representations from streaming video using dictionaries of spatiotemporal features optimized in an unsupervised manner for sparse reconstruction. Non-spiking LCA has previously been used to achieve unsupervised learning of spatiotemporal dictionaries composed of convolutional kernels from raw, unlabeled video. We demonstrate how unsupervised dictionary learning with spiking LCA (S-LCA) can be efficiently implemented using accumulator neurons, which combine a conventional leaky-integrate-and-fire (LIF) spike generator with an additional state variable that is used to minimize the difference between the integrated input and the spiking output. We demonstrate dictionary learning across a wide range of dynamical regimes, from graded to intermittent spiking, for inferring sparse representations of both static images drawn from the CIFAR database as well as video frames captured from a DVS camera. On a classification task that requires identification of the suite from a deck of cards being rapidly flipped through as viewed by a DVS camera, we find essentially no degradation in performance as the LCA model used to infer sparse spatiotemporal representations migrates from graded to spiking. We conclude that accumulator neurons are likely to provide a powerful enabling component of

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Accumulator Neuronsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Dictionary Learning with Accumulator Neurons

Parpart¹,

González²,

Stewart³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…For a long time restricted to small datasets, spiking neural networks now show their strengths when trained directly on temporal data. Future works would include the implementation of these spiking neural networks on a low-power neuromorphic hardware [34], [2], which will enable power-efficient embedded applications.…”

Section: Discussionmentioning

confidence: 99%

“…They are therefore particularly adapted to the processing of event data. Due to the nature of their operations, they are also hardware friendly: previous works showed that on specialized hardware, spiking neural networks consume 50% less energy than traditional neural networks while maintaining the same accuracy [1], and the recent release of new neuromorphic hardware such as Intel Loihi 2 [2] could further improve these results.…”

Section: Introductionmentioning

confidence: 99%

Object Detection with Spiking Neural Networks on Automotive Event Data

Cordone¹,

Miramond²,

Thierion³

2022

Preprint

View full text Add to dashboard Cite

Automotive embedded algorithms have very high constraints in terms of latency, accuracy and power consumption. In this work, we propose to train spiking neural networks (SNNs) directly on data coming from event cameras to design fast and efficient automotive embedded applications. Indeed, SNNs are more biologically realistic neural networks where neurons communicate using discrete and asynchronous spikes, a naturally energy-efficient and hardware friendly operating mode. Event data, which are binary and sparse in space and time, are therefore the ideal input for spiking neural networks. But to date, their performance was insufficient for automotive real-world problems, such as detecting complex objects in an uncontrolled environment. To address this issue, we took advantage of the latest advancements in matter of spike backpropagation -surrogate gradient learning, parametric LIF, SpikingJelly frameworkand of our new voxel cube event encoding to train 4 different SNNs based on popular deep learning networks: SqueezeNet, VGG, MobileNet, and DenseNet. As a result, we managed to increase the size and the complexity of SNNs usually considered in the literature. In this paper, we conducted experiments on two automotive event datasets, establishing new state-of-the-art classification results for spiking neural networks. Based on these results, we combined our SNNs with SSD to propose the first spiking neural networks capable of performing object detection on the complex GEN1 Automotive Detection event dataset.

show abstract