Pith. sign in

REVIEW 3 major objections 5 minor 65 references

Graph-Based Object Classification for Neuromorphic Vision Sensing

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Representing neuromorphic spike streams as a radius-neighborhood graph and classifying them with residual graph CNNs yields top-1 accuracy matching or beating prior event-based methods on six datasets while using less computation and…

desk verdict Event-camera paper with a genuinely useful graph representation and the ASL-DVS dataset, but the state-of-the-art claim rests on accuracy gaps that are within run-to-run noise. read the letter →

arxiv 1908.06648 v1 pith:7WP6HQ44 submitted 2019-08-19 cs.CV

Yin Bi , Aaron Chadha , Alhabib Abbas , Eirina Bourtsoulatze , Yiannis Andreopoulos This is my paper · ORCID
classification cs.CV
keywords neuromorphicvisionsensingeventcamerasgraphconvolutionalnetworksresiduallearningobjectclassificationnon-uniformsamplingradius-neighborhoodAmericanSignLanguagedataset
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Event cameras emit asynchronous spikes rather than frames, which has kept them out of reach of standard convolutional networks. The authors set out to show that a compact graph built from those spikes—each sampled event becomes a node, and nearby events are connected by a weighted spatio-temporal distance—can be classified by a residual graph CNN with B-spline kernels. They report top-1 accuracy that matches or outperforms prior event-based methods on six datasets and beats frame-based deep CNNs such as ResNet50 on five of them, while using about a fifth of the computation of ResNet50. If this holds, neuromorphic sensing gains a direct, gradient-trained path to deep classification without converting spikes to frames. The paper also contributes a new 24-class, 100,800-sample real-world sign-language recording to support further work.

What carries the argument

The central object is the radius-neighborhood graph built from sampled spike events. Each node is an event with coordinates $(x_i,y_i,t_i)$ and polarity feature $p_i \in \{+1,-1\}$; edges connect nodes whose weighted spatio-temporal distance $d_{i,j} = \sqrt{\alpha(|x_i-x_j|^2+|y_i-y_j|^2)+\beta|t_i-t_j|^2}$ is at most $R$, with maximum degree $D_{\max}$. Graph convolution uses a B-spline kernel $g_l(u)$ over pseudo-coordinates $u(i,j) = [|x_i-x_j|, |y_i-y_j|]$ to aggregate neighbor features, followed by batch normalization and ReLU; residual graph blocks add a kernel-size-one shortcut to the main convolution, and max pooling over coordinate clusters coarsens the graph. This machinery preserves the spatio-temporal geometry of spikes while letting standard gradient training work.

What would settle it

Compare per-class accuracy on the released ASL-DVS data with block sizes $k=1$ and $k=8$: if the visually similar letter pairs lose accuracy disproportionately while well-separated letters hold, the random sampling has removed the discriminative cues the method relies on.

Watch

Extended reading notes

Core claim

At the center of the paper is a representation: take a 30 ms window of spikes, discard all but one event from each space-time block of up to $k=8$ events, then treat the survivors as nodes of a radius-neighborhood graph with edge weight $d_{i,j}$ and polarity as the node feature. A stack of residual graph convolution layers with B-spline kernels, batch normalization, max pooling over coordinate clusters, and two fully connected layers maps the graph to a class label. The reported RG-CNN accuracies are $0.990$ on N-MNIST, $0.986$ on MNIST-DVS, $0.657$ on N-Caltech101, $0.540$ on CIFAR10-DVS, $0.914$ on N-CARS, and $0.901$ on the new ASL-DVS dataset; by the authors' accounting this is a new state of the art on five datasets and within $0.001$ of the best published result on N-MNIST. The cost is small: on N-Caltech101 the input graph averages roughly 1,000 nodes against 86,400 pixels in the two-channel event image, and the whole RG-CNN runs at 0.79 GFLOPs with 19.46 MB of parameters.

Load-bearing premise

The load-bearing premise is that randomly keeping just one event from each space-time block of up to eight spikes retains enough of the motion detail that separates classes.

Editorial extensions

If this is right

  • Event cameras could be used directly with gradient-trained deep networks on raw spike streams, removing the need to aggregate spikes into frames.
  • The compact graph input means inference on high-resolution sensors may stay cheap: the paper reports roughly 1,000 graph nodes versus 86,400 frame pixels for N-Caltech101.
  • The reported accuracy on five of six datasets suggests graph-based methods, not frame conversion or spiking networks, may be the most practical route for neuromorphic classification.
  • The released 24-class, 100,800-sample ASL-DVS recording gives the field a larger real-world benchmark to train and compare event-based classifiers.
  • If the accuracy-versus-compression tradeoff holds, the approach can scale to long or dense event streams by sampling more aggressively without retraining the architecture.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper leaves implicit that the same graph construction could transfer to other event-stream tasks, such as detection or tracking, because the graph is defined by spatio-temporal distance rather than a fixed pixel grid.
  • A natural testable extension is a content-aware sampling rule: instead of picking one event at random per space-time block, the selector could favor motion boundaries, potentially recovering the k=1 accuracy at k=8 cost.
  • Because FLOPs depend on node and edge counts rather than image resolution, the approach may scale more gracefully to high-resolution neuromorphic sensors than frame-based CNNs, which the paper does not directly demonstrate.
  • The 30-ms window and k=8 settings were tuned on N-Caltech101; applying the pipeline to longer recordings would require a temporal segmentation strategy that the paper does not specify.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a graph-based representation for neuromorphic vision sensor (NVS) event streams and uses residual graph convolutional networks (RG-CNNs) for object classification. Events are first compressed by non-uniform sampling, then mapped to a radius-neighborhood graph with weighted spatio-temporal distances, and processed by spatial graph convolutions with B-spline kernels, cluster pooling, residual blocks, and fully connected layers. The authors evaluate on N-MNIST, MNIST-DVS, N-Caltech101, CIFAR10-DVS, and N-CARS, and introduce a new 100k-sample, 24-class real-world dataset, ASL-DVS. They report top-1 accuracies that match or exceed prior event-based methods and frame-based deep CNNs, with substantially lower FLOPs and model sizes.

Significance. The graph-based representation is a natural and potentially valuable bridge between asynchronous event data and gradient-trained deep networks, and the paper's complexity analysis in Table 3 makes a concrete quantitative case for the approach. The release of ASL-DVS, described as one of the largest labeled NVS datasets under real-world conditions, is a useful community contribution. If the reported accuracy advantages withstand repeated evaluation, the method would be a meaningful step for neuromorphic object classification. However, the headline state-of-the-art claim is not yet statistically supported: the margins over HATS are close to run-to-run noise and key hyperparameters were selected on the same N-Caltech101 test set used for the main comparison. With proper repeated-split, multi-seed reporting, the paper's central claim could be substantiated; as written, the evidence supports 'competitive with HATS at lower computational cost' more strongly than 'new state-of-the-art.'

major comments (3)
  1. [§5.1, Table 1] The headline state-of-the-art claim is not supported by the reported statistics. The decisive margins over HATS are 1.5 points on N-Caltech101 (0.657 vs 0.642), 1.6 points on CIFAR10-DVS (0.540 vs 0.524), and 1.2 points on N-CARS (0.914 vs 0.902). For test sets of the size used here, the standard error of a proportion is roughly one percentage point, and the paper reports no repeated random splits, no multiple training seeds, and no confidence intervals. The evidence as reported cannot distinguish RG-CNN from HATS on these datasets. Please report mean and standard deviation (or confidence intervals) over multiple splits and seeds and temper the 'new state-of-the-art' wording accordingly.
  2. [§7 (Supplementary), Tables 4–8, and §5.1] Several key hyperparameters (k, R, time-window length, network depth, and kernel size) are selected by maximizing accuracy on N-Caltech101 in the supplementary material, and Table 1 then reports N-Caltech101 accuracy using the selected values. This is selection on the test set and is likely to bias the reported N-Caltech101 number optimistically. The supplementary text itself states that all ablation experiments were conducted on N-Caltech101, so the issue is self-acknowledged. Please move hyperparameter selection to a validation split that is never used for the final comparison, or fix hyperparameters before any test labels are used, and disclose the selection procedure.
  3. [§5.1, Table 1] The text states that RG-CNNs 'consistently outperform' prior methods and 'set a new state-of-the-art' on five of six datasets, but Table 1 shows RG-CNN at 0.990 versus HATS at 0.991 on N-MNIST, and ASL-DVS has no HATS entry. This wording overstates the results. The claim should be corrected to 'matches or exceeds' or the comparison set should be specified precisely.
minor comments (5)
  1. [§3.1 and §3.2] The graph is introduced as directed, but the radius-neighborhood construction yields symmetric edges and the convolution definition in Eq. (3) does not use edge direction. Please clarify whether edges are directed and, if so, how direction enters the aggregation.
  2. [§3.4, Eq. (8)] The fully connected layer formula is not written correctly; the weight tensor F^{P×Min×Q} lacks explicit indices and the feature notation f^p_l is ambiguous. Please rewrite with explicit summation indices.
  3. [Table 3 and Table 9] Inception V4 is reported as 12.25 GFLOPs in Table 3 but 9.24 GFLOPs in Table 9 for the same 224×224 input; please harmonize the FLOPs accounting or explain the difference.
  4. [§5.2] The statement that frame-based CNNs are below state-of-the-art 'since event images contain far less information' is a causal claim not established by the experiments; please rephrase it as an observation rather than an explanation.
  5. [Tables 1 and 2] There are typos in the table headings: 'acccuracy' should be 'accuracy' and 'RG-CNNNs' should be 'RG-CNNs'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the central claims are empirical and benchmarked against external datasets and previously published methods.

full rationale

The paper's derivation chain is not equation-level circular. RG-CNN accuracy is obtained by training on six publicly available NVS datasets and compared with published numbers for HATS, HOTS, H-First, SNN and with frame-based CNNs retrained by the authors (Tables 1-2); the graph construction uses non-uniform grid sampling from Lee et al. [31], SplineCNN B-spline kernels from Fey et al. [22], and residual connections from He et al. [24], all external. The only self-citation, Pix2NVS [6], is cited as an example of emulator-based dataset generation and is not load-bearing for the classification result. Hyperparameters (k, R, depth, kernel size) are tuned in the supplementary on N-Caltech101, so reporting N-Caltech101 accuracy with those choices is a potential selection-bias issue, but it is not circular: the reported numbers are measured outcomes, not quantities defined to equal the tuning criterion. There is no fitted parameter renamed as a prediction, no uniqueness theorem, and no ansatz smuggled in via self-citation.

Assumptions & free parameters 9 free parameters · 4 assumptions · 0 invented entities

The central method depends on a set of hyperparameters selected on N-Caltech101 and applied across datasets, plus the domain assumptions that event streams live in a Euclidean spatio-temporal space and that polarity alone is a sufficient node signal. No new physical entities are postulated.

free parameters (9)
  • alpha (spatial weight) = 1
    Weights spatial coordinate differences in Eq (2). Chosen by hand to balance spatial and temporal resolutions.
  • beta (temporal weight) = 0.5e-5
    Weights timestamp differences in Eq (2). Chosen by hand to compensate for the higher timing accuracy of NVS.
  • R (radius distance) = 3
    Neighborhood radius in Eq (2). Selected via ablation on N-Caltech101 (Table 5).
  • Dmax (max degree) = 32
    Limits graph size by capping connectivity per node. Set by hand.
  • k (sampling grid occupancy) = 8
    Maximum events per space-time volume in non-uniform sampling; one event is randomly selected per volume. Selected via ablation on N-Caltech101 (Table 4).
  • time window = 30 ms
    Random 30ms event window extracted per sample. Selected via ablation on N-Caltech101 (Table 6).
  • kernel size = 5
    B-spline kernel size per dimension. Selected via ablation on N-Caltech101 (Table 8).
  • network depth = 2 blocks for N-MNIST and MNIST-DVS, 3 blocks for others, 4 layers for N-Caltech101
    Number of residual blocks chosen per dataset complexity; for N-Caltech101 depth 4 was chosen via ablation (Table 7).
  • pooling cluster sizes = per-dataset values
    Chosen by hand to match sensor resolution and pooling schedule.
assumptions (4)
  • domain assumption NVS events can be embedded as points in spatio-temporal Euclidean space with distance Eq (2)
    The graph construction assumes that a weighted Euclidean distance between (x,y,t) coordinates captures the relevant relation between spikes. This is introduced in Section 3.1.
  • domain assumption Event polarity alone is a sufficient initial node feature
    The input feature f(0)(i) is set to p_i in Section 3.1; no intensity, timestamp, or local density is used as input.
  • domain assumption Previously published baseline accuracies (for example, HATS in [57]) are reliable and comparable
    The state-of-the-art comparison in Table 1 relies on numbers reported by Sironi et al. under their own protocol; comparability is assumed.
  • domain assumption Randomly sampling one event per space-time volume retains enough information
    The compression step in Section 3.1 assumes the sampled subset preserves class-discriminative structure; Table 4 tests but does not prove this beyond the 30ms windows used.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph-Based Object Classification for Neuromorphic Vision Sensing." pith.science (2026). https://pith.science/paper/7WP6HQ44

@misc{pith2026190806648,
  author       = {Pith},
  title        = {Pith review of: Graph-Based Object Classification for Neuromorphic Vision Sensing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7WP6HQ44}},
  note         = {Machine review of arXiv:1908.06648}
}
read the original abstract

Neuromorphic vision sensing (NVS)\ devices represent visual information as sequences of asynchronous discrete events (a.k.a., ``spikes'') in response to changes in scene reflectance. Unlike conventional active pixel sensing (APS), NVS allows for significantly higher event sampling rates at substantially increased energy efficiency and robustness to illumination changes. However, object classification with NVS streams cannot leverage on state-of-the-art convolutional neural networks (CNNs), since NVS does not produce frame representations. To circumvent this mismatch between sensing and processing with CNNs, we propose a compact graph representation for NVS. We couple this with novel residual graph CNN architectures and show that, when trained on spatio-temporal NVS data for object classification, such residual graph CNNs preserve the spatial and temporal coherence of spike events, while requiring less computation and memory. Finally, to address the absence of large real-world NVS datasets for complex recognition tasks, we present and make available a 100k dataset of NVS recordings of the American sign language letters, acquired with an iniLabs DAVIS240c device under real-world conditions.

Figures

Figures reproduced from arXiv: 1908.06648 by the authors.

Figure 1
Figure 1. Examples of objects captured by APS and neu [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Framework of graph-based object classification fo [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Examples of the ASL-DVS dataset (the visual [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Comparison of proposed NVS dataset w.r.t. the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 61 canonical work pages

  1. [1]

    Truenorth: Design and tool flow of a 65 mw 1 million neuron pro- grammable neurosynaptic chip

    Filipp Akopyan, Jun Sawada, Andrew Cassidy, Rodrigo Alvarez-Icaza, John Arthur, Paul Merolla, Nabil Imam, Y u- taka Nakamura, Pallab Datta, Gi-Joon Nam, et al. Truenorth: Design and tool flow of a 65 mw 1 million neuron pro- grammable neurosynaptic chip. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems , 34(10):1537–1557, 2015. 2

    work page 2015

  2. [2]

    Diffusion-convolutional neural networks

    James Atwood and Don Towsley. Diffusion-convolutional neural networks. In Advances in Neural Information Pro- cessing Systems, pages 1993–2001, 2016. 3

    work page 1993

  3. [3]

    Contour motion estimation for asynchronous event- driven cameras

    Francisco Barranco, Cornelia Ferm¨ uller, and Yiannis A loi- monos. Contour motion estimation for asynchronous event- driven cameras. Proceedings of the IEEE , 102(10):1537– 1556, 2014. 2

    work page 2014

  4. [4]

    Bio-inspired motion estimation with event-driven sensors

    Francisco Barranco, Cornelia Fermuller, and Yiannis Al oi- monos. Bio-inspired motion estimation with event-driven sensors. In International W ork-Conference on Artificial Neu- ral Networks, pages 309–321. Springer, 2015. 2

    work page 2015

  5. [5]

    Event-based visual flow

    Ryad Benosman, Charles Clercq, Xavier Lagorce, Sio-Hoi Ieng, and Chiara Bartolozzi. Event-based visual flow. IEEE transactions on neural networks and learning systems , 25(2):407–417, 2014. 2

    work page 2014

  6. [6]

    Pix2nvs: Parameterize d conversion of pixel-domain video frames to neuromorphic vision streams

    Yin Bi and Yiannis Andreopoulos. Pix2nvs: Parameterize d conversion of pixel-domain video frames to neuromorphic vision streams. In 2017 IEEE International Conference on Image Processing (ICIP), pages 1990–1994. IEEE, 2017. 2, 6

    work page 2017

  7. [7]

    Extraction of temporally correlated features from dynamic vision sensor s with spike-timing-dependent plasticity

    Olivier Bichler, Damien Querlioz, Simon J Thorpe, Jean- Philippe Bourgoin, and Christian Gamrat. Extraction of temporally correlated features from dynamic vision sensor s with spike-timing-dependent plasticity. Neural Networks , 32:339–348, 2012. 2

    work page 2012

  8. [8]

    Learning shape correspondence with anisotropic convolutional neural networks

    Davide Boscaini, Jonathan Masci, Emanuele Rodol` a, and Michael Bronstein. Learning shape correspondence with anisotropic convolutional neural networks. In Advances in Neural Information Processing Systems , pages 3189–3197, 2016. 3, 4

    work page 2016

Show all 65 references
  1. [9]

    Geometric deep learning: going beyond euclidean data

    Michael M Bronstein, Joan Bruna, Yann LeCun, Arthur Szlam, and Pierre V andergheynst. Geometric deep learning: going beyond euclidean data. IEEE Signal Processing Mag- azine, 34(4):18–42, 2017. 4

    2017

  2. [10]

    On event-based optical flow detection

    Tobias Brosch, Stephan Tschechne, and Heiko Neumann. On event-based optical flow detection. Frontiers in neuro- science, 9:137, 2015. 2

    2015

  3. [11]

    Spectral networks and locally connected networks on graphs

    Joan Bruna, Wojciech Zaremba, Arthur Szlam, and Yann Le - Cun. Spectral networks and locally connected networks on graphs. arXiv preprint arXiv:1312.6203, 2013. 3, 4

    2013 arXiv

  4. [12]

    Attention mechanisms for object recognition with event-based cameras

    Marco Cannici, Marco Ciccone, Andrea Romanoni, and Matteo Matteucci. Attention mechanisms for object recognition with event-based cameras. arXiv preprint arXiv:1807.09480, 2018. 1, 2

    2018 arXiv

  5. [13]

    Event-based convolutional networks for object detection in neuromorphic cameras

    Marco Cannici, Marco Ciccone, Andrea Romanoni, and Matteo Matteucci. Event-based convolutional networks for object detection in neuromorphic cameras. arXiv preprint arXiv:1805.07931, 2018. 1, 2

    2018 arXiv

  6. [14]

    Asyn- chronous event-based corner detection and matching

    Xavier Clady, Sio-Hoi Ieng, and Ryad Benosman. Asyn- chronous event-based corner detection and matching. Neural Networks, 66:91–106, 2015. 2

    2015

  7. [15]

    A motion-based feature for event-based pattern recognition

    Xavier Clady, Jean-Matthieu Maro, S´ ebastien Barr´ e, and Ryad B Benosman. A motion-based feature for event-based pattern recognition. Frontiers in neuroscience, 10:594, 2017. 1, 2

    2017

  8. [16]

    Event-based feature detection, recog- nition and classification

    Gregory Kevin Cohen. Event-based feature detection, recog- nition and classification . PhD thesis, Paris 6, 2016. 1, 2

    2016

  9. [17]

    Convolutional neural networks on graphs with fast localized spectral filtering

    Micha¨ el Defferrard, Xavier Bresson, and Pierre V an- dergheynst. Convolutional neural networks on graphs with fast localized spectral filtering. In Advances in neural infor- mation processing systems , pages 3844–3852, 2016. 3, 4, 7

    2016

  10. [18]

    Neuromorphic vision sensing and proces s- ing

    Tobi Delbruck. Neuromorphic vision sensing and proces s- ing. In Europ. Solid-State Dev. Res. Conf, pages 7–14, 2016. 1

    2016

  11. [19]

    Unsupervised learning of digit recognition using spike-timing-dependent plasti city

    Peter U Diehl and Matthew Cook. Unsupervised learning of digit recognition using spike-timing-dependent plasti city. Frontiers in computational neuroscience, 9:99, 2015. 2

    2015

  12. [20]

    Fast-classifying, hi gh- accuracy spiking deep networks through weight and thresh- old balancing

    Peter U Diehl, Daniel Neil, Jonathan Binas, Matthew Coo k, Shih-Chii Liu, and Michael Pfeiffer. Fast-classifying, hi gh- accuracy spiking deep networks through weight and thresh- old balancing. In 2015 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE, 2015. 3

    2015

  13. [21]

    Convolutional networks on graphs for learn- ing molecular fingerprints

    David K Duvenaud, Dougal Maclaurin, Jorge Iparraguirr e, Rafael Bombarell, Timothy Hirzel, Al´ an Aspuru-Guzik, and Ryan P Adams. Convolutional networks on graphs for learn- ing molecular fingerprints. In Advances in neural informa- tion processing systems, pages 2224–2232, 2015. 3

    2015

  14. [22]

    Splinecnn: Fast geometric deep learning with continuous b-spline kernels

    Matthias Fey, Jan Eric Lenssen, Frank Weichert, and Hei n- rich M¨ uller. Splinecnn: Fast geometric deep learning with continuous b-spline kernels. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 869–877, 2018. 3, 4, 8, 10

    2018

  15. [23]

    pydvs: An extensible, real-time dynamic vision sensor emulator using off-the-shelf hardware

    Garibaldi Pineda Garc´ ıa, Patrick Camilleri, Qian Liu , and Steve Furber. pydvs: An extensible, real-time dynamic vision sensor emulator using off-the-shelf hardware. In 2016 IEEE Symposium Series on Computational Intelligence (SSCI), pages 1–7. IEEE, 2016. 2, 6

    2016

  16. [24]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 1, 2, 5, 7, 8

    2016

  17. [25]

    Dvs benchmark datasets for object tracking, action recognition, and object recognition

    Y uhuang Hu, Hongjie Liu, Michael Pfeiffer, and Tobi Del - bruck. Dvs benchmark datasets for object tracking, action recognition, and object recognition. Frontiers in neuro- science, 10:405, 2016. 2

    2016

  18. [26]

    Batch normalizati on: Accelerating deep network training by reducing internal co - variate shift

    Sergey Ioffe and Christian Szegedy. Batch normalizati on: Accelerating deep network training by reducing internal co - variate shift. ICML, 2015. 5

    2015

  19. [27]

    Semi-supervised classi- fication with graph convolutional networks

    Thomas N Kipf and Max Welling. Semi-supervised classi- fication with graph convolutional networks. ICLR, 2017. 3, 7

    2017

  20. [28]

    Imagenet classification with deep convolutional neural net - works

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton . Imagenet classification with deep convolutional neural net - works. In Advances in neural information processing sys- tems, pages 1097–1105, 2012. 1

    2012

  21. [29]

    Hots: a hierarchy of event-based time-surfaces for pattern recognition

    Xavier Lagorce, Garrick Orchard, Francesco Galluppi, Bertram E Shi, and Ryad B Benosman. Hots: a hierarchy of event-based time-surfaces for pattern recognition. IEEE transactions on pattern analysis and machine intelligence , 39(7):1346–1359, 2017. 1, 2, 7

    2017

  22. [30]

    Tra in- ing deep spiking neural networks using backpropagation

    Jun Haeng Lee, Tobi Delbruck, and Michael Pfeiffer. Tra in- ing deep spiking neural networks using backpropagation. Frontiers in neuroscience, 10:508, 2016. 2, 7

    2016

  23. [31]

    Point data reduction using 3d grids

    KH Lee, H Woo, and T Suk. Point data reduction using 3d grids. The International Journal of Advanced Manufacturing Technology, 18(3):201–210, 2001. 4

    2001

  24. [32]

    Cayleynets: Graph convolutional neural networks with complex rational spectral filters

    Ron Levie, Federico Monti, Xavier Bresson, and Michael M Bronstein. Cayleynets: Graph convolutional neural networks with complex rational spectral filters. IEEE Transactions on Signal Processing, 67(1):97–109, 2017. 3

    2017

  25. [33]

    Cifar10-dvs: an event-stream dataset for ob- ject classification

    Hongmin Li, Hanchao Liu, Xiangyang Ji, Guoqi Li, and Luping Shi. Cifar10-dvs: an event-stream dataset for ob- ject classification. Frontiers in neuroscience, 11:309, 2017. 2, 6

    2017

  26. [34]

    Deeper insight s into graph convolutional networks for semi-supervised learn- ing

    Qimai Li, Zhichao Han, and Xiao-Ming Wu. Deeper insight s into graph convolutional networks for semi-supervised learn- ing. In Thirty-Second AAAI Conference on Artificial Intelli- gence, 2018. 5, 9

    2018

  27. [35]

    A 128x128, 120 db 30mw latency asynchronous temporal contrast vision sensor

    Patrick Lichtsteiner, Christoph Posch, and Tobi Delbr uck. A 128x128, 120 db 30mw latency asynchronous temporal contrast vision sensor. IEEE journal of solid-state circuits , 43(2):566–576, 2008. 6

    2008

  28. [36]

    Geodesic convolutional neural net- works on riemannian manifolds

    Jonathan Masci, Davide Boscaini, Michael Bronstein, a nd Pierre V andergheynst. Geodesic convolutional neural net- works on riemannian manifolds. In Proceedings of the IEEE international conference on computer vision work- shops, pages 37–45, 2015. 3, 4

    2015

  29. [37]

    Pruning convolutional neural networks for resource efficient inference

    Pavlo Molchanov, Stephen Tyree, Tero Karras, Timo Aila , and Jan Kautz. Pruning convolutional neural networks for resource efficient inference. ICLR, 2017. 8

    2017

  30. [38]

    Geometric deep learning on graphs and manifolds using mixture model cnns

    Federico Monti, Davide Boscaini, Jonathan Masci, Emanuele Rodola, Jan Svoboda, and Michael M Bronstein. Geometric deep learning on graphs and manifolds using mixture model cnns. In Proceedings of the IEEE Confer- ence on Computer Vision and Pattern Recognition , pages 5115–512...

    2017

  31. [39]

    Fast event-based corner detection

    Elias Mueggler, Chiara Bartolozzi, and Davide Scaramu zza. Fast event-based corner detection. In British Machine Vis. Conf.(BMVC), volume 1, 2017. 2

    2017

  32. [40]

    The event-camera dataset and simulator: Event-based data for pose estimation, visua l odometry, and slam

    Elias Mueggler, Henri Rebecq, Guillermo Gallego, Tobi Del- bruck, and Davide Scaramuzza. The event-camera dataset and simulator: Event-based data for pose estimation, visua l odometry, and slam. The International Journal of Robotics Research, 36(2):142–149, 2017. 2, 6

    2017

  33. [41]

    Event-driven contrastive divergence for spiking neuromorphic systems

    Emre Neftci, Srinjoy Das, Bruno Pedroni, Kenneth Kreut z- Delgado, and Gert Cauwenberghs. Event-driven contrastive divergence for spiking neuromorphic systems. Frontiers in neuroscience, 7:272, 2014. 2

    2014

  34. [42]

    Neuromorph ic engineering

    E Neftci, C Posch, E Chicca, and H Ishibuchi. Neuromorph ic engineering. Computational Intelligence, 2:278, 2015. 1

    2015

  35. [43]

    Phas ed lstm: Accelerating recurrent network training for long or event-based sequences

    Daniel Neil, Michael Pfeiffer, and Shih-Chii Liu. Phas ed lstm: Accelerating recurrent network training for long or event-based sequences. In Advances in Neural Information Processing Systems, pages 3882–3890, 2016. 7

    2016

  36. [44]

    Real-time classification and sensor f u- sion with a spiking deep belief network

    Peter O’Connor, Daniel Neil, Shih-Chii Liu, Tobi Delbr uck, and Michael Pfeiffer. Real-time classification and sensor f u- sion with a spiking deep belief network. Frontiers in neuro- science, 7:178, 2013. 3

    2013

  37. [45]

    Bioin- spired visual motion estimation

    Garrick Orchard and Ralph Etienne-Cummings. Bioin- spired visual motion estimation. Proceedings of the IEEE , 102(10):1520–1536, 2014. 2

    2014

  38. [46]

    Converting static image datasets to spiking neuromorphic datasets using saccades

    Garrick Orchard, Ajinkya Jayawant, Gregory K Cohen, an d Nitish Thakor. Converting static image datasets to spiking neuromorphic datasets using saccades. Frontiers in neuro- science, 9:437, 2015. 6

    2015

  39. [47]

    Hfirst: a temporal approach to object recognition

    Garrick Orchard, Cedric Meyer, Ralph Etienne-Cumming s, Christoph Posch, Nitish Thakor, and Ryad Benosman. Hfirst: a temporal approach to object recognition. IEEE transactions on pattern analysis and machine intelligence , 37(10):2028–2040, 2015. 7

    2015

  40. [48]

    Bag of events: an efficient probability-based feature extracti on method for aer image sensors

    Xi Peng, Bo Zhao, Rui Yan, Huajin Tang, and Zhang Yi. Bag of events: an efficient probability-based feature extracti on method for aer image sensors. IEEE transactions on neural networks and learning systems , 28(4):791–803, 2017. 2

    2017

  41. [49]

    Mapping from frame-driven to frame-free event-driven vision systems by low-rate rate coding and coincidence processing–application to feedfor - ward convnets

    Jos´ e Antonio P´ erez-Carrasco, Bo Zhao, Carmen Serran o, Begona Acha, Teresa Serrano-Gotarredona, Shouchun Chen, and Bernab´ e Linares-Barranco. Mapping from frame-driven to frame-free event-driven vision systems by low-rate rate coding and coincidence processing–application...

    2013

  42. [50]

    The NURBS book

    Les Piegl and Wayne Tiller. The NURBS book . Springer Science & Business Media, 2012. 4

    2012

  43. [51]

    A qvga 143 db dynamic range frame-free pwm image sensor with lossless pixel-level video compression and time-domain cds

    Christoph Posch, Daniel Matolin, and Rainer Wohlgenan nt. A qvga 143 db dynamic range frame-free pwm image sensor with lossless pixel-level video compression and time-domain cds. IEEE Journal of Solid-State Circuits , 46(1):259–275, 2011. 1, 6

    2011

  44. [52]

    Discrete sign al processing on graphs

    Aliaksei Sandryhaila and Jos´ e MF Moura. Discrete sign al processing on graphs. IEEE transactions on signal process- ing, 61(7):1644–1656, 2013. 3, 4

    2013

  45. [53]

    Poker-dvs and mnist-dvs

    Teresa Serrano-Gotarredona and Bernab´ e Linares-Barranco. Poker-dvs and mnist-dvs. their history, how they were made, and other details. Frontiers in neuroscience, 9:481, 2015. 2, 6

    2015

  46. [54]

    The emerging field of signal processing on graphs: Extending high-dimensiona l data analysis to networks and other irregular domains

    David I Shuman, Sunil K Narang, Pascal Frossard, Anto- nio Ortega, and Pierre V andergheynst. The emerging field of signal processing on graphs: Extending high-dimensiona l data analysis to networks and other irregular domains. arXiv preprint arXiv:1211.0053, 2012. 3, 4

    2012 arXiv

  47. [55]

    Dynamic edge- conditioned filters in convolutional neural networks on graphs

    Martin Simonovsky and Nikos Komodakis. Dynamic edge- conditioned filters in convolutional neural networks on graphs. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 3693–3702, 2017. 5

    2017

  48. [56]

    V ery deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. V ery deep convo- lutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014. 7, 8

    2014 arXiv

  49. [57]

    Hats: histograms of aver- aged time surfaces for robust event-based object classifica - tion

    Amos Sironi, Manuele Brambilla, Nicolas Bourdis, Xavi er Lagorce, and Ryad Benosman. Hats: histograms of aver- aged time surfaces for robust event-based object classifica - tion. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1731–1740, ...

    2018

  50. [58]

    Scal- able energy-efficient, low-latency implementations of tra ined spiking deep belief networks on spinnaker

    Evangelos Stromatias, Daniel Neil, Francesco Gallupp i, Michael Pfeiffer, Shih-Chii Liu, and Steve Furber. Scal- able energy-efficient, low-latency implementations of tra ined spiking deep belief networks on spinnaker. In 2015 Interna- tional Joint Conference on Neural Networks...

    2015

  51. [59]

    Inception-v4, inception-resnet and the impact of residual connections on learning

    Christian Szegedy, Sergey Ioffe, Vincent V anhoucke, a nd Alexander A Alemi. Inception-v4, inception-resnet and the impact of residual connections on learning. In Thirty-First AAAI Conference on Artificial Intelligence , 2017. 7, 8

    2017

  52. [60]

    Ben ch- marking neuromorphic vision: lessons learnt from computer vision

    Cheston Tan, Stephane Lallee, and Garrick Orchard. Ben ch- marking neuromorphic vision: lessons learnt from computer vision. Frontiers in neuroscience, 9:374, 2015. 1

    2015

  53. [61]

    Fast event-based harris corner detection exploiting the advantages of event-driven cameras

    V alentina V asco, Arren Glover, and Chiara Bartolozzi. Fast event-based harris corner detection exploiting the advantages of event-driven cameras. In 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 4144–4149. IEEE, 2016. 2

    2016

  54. [62]

    Local spec- tral graph convolution for point set feature learning

    Chu Wang, Babak Samari, and Kaleem Siddiqi. Local spec- tral graph convolution for point set feature learning. In Pro- ceedings of the European Conference on Computer Vision (ECCV), pages 52–66, 2018. 3, 4

    2018

  55. [63]

    How powerful are graph neural networks? ICLR, 2019

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegel ka. How powerful are graph neural networks? ICLR, 2019. 7

    2019

  56. [64]

    Exploit all the layers: Fast and accurate cnn object detector with scale de- pendent pooling and cascaded rejection classifiers

    Fan Yang, Wongun Choi, and Y uanqing Lin. Exploit all the layers: Fast and accurate cnn object detector with scale de- pendent pooling and cascaded rejection classifiers. In Pro- ceedings of the IEEE conference on computer vision and pat- tern recognition, pages 2129–2137, 2016. 5

    2016

  57. [65]

    Ev-flownet: self-supervised optical flow estimation for event-based cameras

    Alex Zihao Zhu, Liangzhe Y uan, Kenneth Chaney, and Kostas Daniilidis. Ev-flownet: self-supervised optical flow estimation for event-based cameras. arXiv preprint arXiv:1802.06898, 2018. 1, 2, 7

    2018 arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.