pith. sign in

arxiv: 2309.09550 · v4 · submitted 2023-09-18 · 💻 cs.NE · cs.AI

Adaptive Reorganization of Neural Pathways for Continual Learning with Spiking Neural Networks

Bing Han , Feifei Zhao , Wenxuan Pan , Zhuoya Zhao , Xianqi Li , Qingqun Kong , Yi Zeng This is my paper

Pith reviewed 2026-05-24 07:03 UTC · model grok-4.3

classification 💻 cs.NE cs.AI
keywords continual learningspiking neural networksself-organizing regulationneural pathwaysbackward transferself-repairingenergy efficiencyincremental tasks
0
0 comments X

The pith

The SOR-SNN model uses self-organizing regulation to turn one limited spiking neural network into multiple sparse pathways that handle sequences of tasks without performance loss or rising energy use.

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

The paper proposes a brain-inspired continual learning algorithm that employs Self-Organizing Regulation networks to reorganize a single spiking neural network into rich sparse neural pathways. This reorganization is claimed to let the model learn new tasks incrementally while preserving earlier performance, lowering energy consumption, and increasing memory capacity compared with standard approaches. A sympathetic reader would care because the method is presented as mimicking how biological brains manage many cognitive tasks with fixed resources, potentially enabling more efficient artificial systems for sequential learning scenarios. The work further reports backward transfer that improves old tasks and self-repair after damage or pruning.

Core claim

The central claim is that Self-Organizing Regulation networks can reorganize the single and limited Spiking Neural Network (SOR-SNN) into rich sparse neural pathways to efficiently cope with incremental tasks, yielding consistent superiority in performance, energy consumption, and memory capacity across child-like simple to complex tasks as well as on generalized CIFAR100 and ImageNet datasets, while also showing backward transfer that integrates past knowledge with new information and self-repairing ability after irreversible damage.

What carries the argument

Self-Organizing Regulation (SOR) networks that dynamically reorganize a spiking neural network into task-specific sparse pathways.

If this is right

  • The model can learn more complex tasks as well as more tasks overall while maintaining or improving prior performance.
  • Past learned knowledge integrates with current-task information, producing backward transfer that facilitates old tasks.
  • The model exhibits self-repairing ability after irreversible damage.
  • For pruned networks the model automatically allocates new pathways from the retained network to recover forgotten knowledge.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same reorganization principle could be tested on non-spiking architectures to check whether the energy and capacity gains are specific to spike-based computation.
  • If the sparse pathways remain stable across hundreds of tasks, the approach might reduce the hardware scaling cost of continual learning in embedded systems.
  • Measuring pathway overlap between similar tasks could reveal whether the method implicitly discovers task similarity without explicit supervision.

Load-bearing premise

Self-Organizing Regulation networks can reorganize the single and limited Spiking Neural Network into rich sparse neural pathways to efficiently cope with incremental tasks without performance drop and energy consumption rise.

What would settle it

Running SOR-SNN on a sequence of fifty or more increasingly complex tasks and measuring whether accuracy stays flat or rises, energy per inference stays flat or falls, and memory footprint grows sublinearly compared with fixed-architecture baselines; any consistent degradation in these metrics would falsify the central claim.

Figures

Figures reproduced from arXiv: 2309.09550 by Bing Han, Feifei Zhao, Qingqun Kong, Wenxuan Pan, Xianqi Li, Yi Zeng, Zhuoya Zhao.

Figure 1
Figure 1. Figure 1: Sparse neural pathways self-organized collaboration for continual learning. Purple neurons and cyan neurons are individual neurons for task 1 and task 2, respectively, and blue neurons are shared for both tasks. In the blue box, the different synapses of neuron D are utilized for different tasks and form sparse connections. The human brain could dynamically reorganize neural circuits during continual de￾ve… view at source ↗
Figure 2
Figure 2. Figure 2: The procedure of SOR-SNN model. Each spiking neural network block in the proposed SOR-SNN model involves a self-organizing regulation network which is responsible for selectively activating task-specific sparse pathways in the SNN. For example: the purple connec￾tions form the pathway for task 1. In particular, the self-organizing regulatory network contains the fundamental weighing module and the path sea… view at source ↗
Figure 3
Figure 3. Figure 3: Validation of child-like simple-to-complex continual learning. (A-C) The simple to complex cognitive tasks include sketches, cartoons and photos. (E-G) Task-specific sparse path￾ways, for example, the blue, yellow and purple arrows represent the pathways for Task 1, Task 2 and Task 3 respectively in the fully connected output layer. (D,H,L) Visualization of synaptic acti￾vation counts in partial convolutio… view at source ↗
Figure 4
Figure 4. Figure 4: The comparative performance of SOR-SNN on diverse continual learning tasks. The average accuracy (A-C) and the number of inactive parameters (D-F) of the network for simple to complex cognitive tasks, the CIFAR100 and Mini-ImageNet datasets. The average accuracy on the large scale ImageNet dataset (G). for each cognitive task as shown in [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: (A) The current test accuracy of past learned tasks in our SOR-SNN model. (B-C) The effect of memory loss coefficient and orthogonal loss coefficient on performance. (D) Injury schematic, containing the initial network, the network with task-specific pathways assigned and the network after injury task 1 of SNNs. (E) Accuracy comparisons before and after injury of first four tasks on CIFAR100 10steps. For t… view at source ↗
read the original abstract

The human brain can self-organize rich and diverse sparse neural pathways to incrementally master hundreds of cognitive tasks. However, most existing continual learning algorithms for deep artificial and spiking neural networks are unable to adequately auto-regulate the limited resources in the network, which leads to performance drop along with energy consumption rise as the increase of tasks. In this paper, we propose a brain-inspired continual learning algorithm with adaptive reorganization of neural pathways, which employs Self-Organizing Regulation networks to reorganize the single and limited Spiking Neural Network (SOR-SNN) into rich sparse neural pathways to efficiently cope with incremental tasks. The proposed model demonstrates consistent superiority in performance, energy consumption, and memory capacity on diverse continual learning tasks ranging from child-like simple to complex tasks, as well as on generalized CIFAR100 and ImageNet datasets. In particular, the SOR-SNN model excels at learning more complex tasks as well as more tasks, and is able to integrate the past learned knowledge with the information from the current task, showing the backward transfer ability to facilitate the old tasks. Meanwhile, the proposed model exhibits self-repairing ability to irreversible damage and for pruned networks, could automatically allocate new pathway from the retained network to recover memory for forgotten knowledge.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The manuscript proposes SOR-SNN, a continual learning method for spiking neural networks that employs Self-Organizing Regulation (SOR) networks to dynamically reorganize a single limited SNN into multiple sparse neural pathways. The central claims are that this yields consistent gains in accuracy, energy efficiency, and memory capacity across incremental task sequences (including child-like simple tasks up to complex ones), demonstrates backward transfer that improves prior tasks, and exhibits self-repair after irreversible damage or pruning, with supporting experiments on CIFAR-100 and ImageNet.

Significance. If the empirical results hold under scrutiny, the work offers a concrete mechanism for resource-efficient lifelong learning in SNNs that avoids the typical accuracy-energy trade-off seen in existing continual-learning approaches. The evaluation across a range of task difficulties and standard image-classification benchmarks provides a useful data point for neuromorphic continual-learning research.

minor comments (3)
  1. [Abstract] The abstract asserts performance superiority, backward transfer, and self-repair without any quantitative values, error bars, or dataset names; these claims should be summarized with key numbers already in the abstract so readers can assess strength immediately.
  2. [Figure 3] Figure captions for the pathway-reorganization diagrams do not indicate the number of runs or the precise metric used to color-code active pathways; this reduces interpretability of the visual evidence for adaptive reorganization.
  3. [§5.4] The self-repair experiments on pruned networks report recovery of accuracy but do not state whether the retained sub-network size or the number of new pathways allocated is controlled across compared methods; a controlled ablation would strengthen the claim.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, recognition of the work's potential impact on neuromorphic continual learning, and recommendation of minor revision. No specific major comments were provided in the report.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper presents an empirical continual-learning algorithm (SOR-SNN) whose central claims rest on experimental results across multiple datasets (CIFAR-100, ImageNet, etc.). No derivation chain, fitted-parameter predictions, or self-citation load-bearing steps are described; performance, energy, and memory metrics are reported as measured outcomes rather than derived by construction from the inputs. The architecture description and reorganization mechanism are introduced as design choices whose efficacy is then validated externally, satisfying the criteria for a self-contained empirical result.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no information on free parameters, axioms, or invented entities.

pith-pipeline@v0.9.0 · 5768 in / 1109 out tokens · 28208 ms · 2026-05-24T07:03:17.133500+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

68 extracted references · 68 canonical work pages · 6 internal anchors

  1. [1]

    Lewis, M. D. Self-organizing individual differences in brain development. Devel- opmental Review 25, 252–277 (2005)

    work page 2005

  2. [2]

    Seven properties of self-organization in the human brain

    Dresp-Langley, B. Seven properties of self-organization in the human brain. Big Data and Cognitive Computing 4, 10 (2020)

    work page 2020

  3. [3]

    Tau, G. Z. & Peterson, B. S. Normal development of brain circuits. Neuropsy- chopharmacology 35, 147–168 (2010)

    work page 2010

  4. [4]

    Nelson, C. A. Neural plasticity and human development: The role of early experience in sculpting memory systems. Developmental Science 3, 115–136 (2000)

    work page 2000

  5. [5]

    Huttenlocher, P. R. Neural plasticity: The effects of environment on the devel- opment of the cerebral cortex (Harvard University Press, 2009). 17

    work page 2009

  6. [6]

    & DiCarlo, J

    Kourtzi, Z. & DiCarlo, J. J. Learning and neural plasticity in visual object recognition. Current opinion in neurobiology 16, 152–158 (2006)

    work page 2006

  7. [7]

    & Somers, D

    Grossberg, S. & Somers, D. Synchronized oscillations during cooperative feature linking in a cortical model of visual perception. Neural networks 4, 453–466 (1991)

    work page 1991

  8. [8]

    Lipton, S. A. & Kater, S. B. Neurotransmitter regulation of neuronal outgrowth, plasticity and survival. Trends in neurosciences 12, 265–270 (1989)

    work page 1989

  9. [9]

    Jacobs, K. M. & Donoghue, J. P. Reshaping the cortical motor map by unmask- ing latent intracortical connections. Science 251, 944–947 (1991)

    work page 1991

  10. [10]

    Barron, H. C. Neural inhibition for continual learning and memory. Current opinion in neurobiology 67, 85–94 (2021)

    work page 2021

  11. [11]

    Huttenlocher, P. R. Morphometric study of human cerebral cortex development. Neuropsychologia 28, 517–527 (1990)

    work page 1990

  12. [12]

    Mayford, M., Siegelbaum, S. A. & Kandel, E. R. Synapses and memory storage. Cold Spring Harbor perspectives in biology 4, a005751 (2012)

    work page 2012

  13. [13]

    Kirkpatrick, J. et al. Overcoming catastrophic forgetting in neural networks. Proceedings of the national academy of sciences 114, 3521–3526 (2017)

    work page 2017

  14. [14]

    & Ganguli, S

    Zenke, F., Poole, B. & Ganguli, S. Continual learning through synaptic intel- ligence. In International Conference on Machine Learning , 3987–3995 (PMLR, 2017)

    work page 2017

  15. [15]

    & Tuytelaars, T

    Aljundi, R., Babiloni, F., Elhoseiny, M., Rohrbach, M. & Tuytelaars, T. Memory aware synapses: Learning what (not) to forget. In Proceedings of the European Conference on Computer Vision (ECCV) , 139–154 (2018)

    work page 2018

  16. [16]

    & Hoiem, D

    Li, Z. & Hoiem, D. Learning without forgetting. IEEE transactions on pattern analysis and machine intelligence 40, 2935–2947 (2017)

    work page 2017

  17. [17]

    & Ohara, K

    Murata, K., Ito, S. & Ohara, K. Learning and transforming general represen- tations to break down stability-plasticity dilemma. In Proceedings of the Asian Conference on Computer Vision , 3994–4010 (2022)

    work page 2022

  18. [18]

    Von Oswald, J., Henning, C., Grewe, B. F. & Sacramento, J. Continual learning with hypernetworks. arXiv preprint arXiv:1906.00695 (2019)

    work page arXiv 1906

  19. [19]

    S., Varshney, S., Srijith, P

    Chandra, D. S., Varshney, S., Srijith, P. & Gupta, S. Continual learning with dependency preserving hypernetworks. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2339–2348 (2023)

    work page 2023

  20. [20]

    Beaulieu, S. et al. Learning to continually learn.arXiv preprint arXiv:2002.09571 (2020). 18

    work page arXiv 2002

  21. [21]

    & Steven, H

    Pham, Q., Liu, C., Sahoo, D. & Steven, H. Contextual transformation net- works for online continual learning. In International Conference on Learning Representations (2020)

    work page 2020

  22. [22]

    Rusu, A. A. et al. Progressive neural networks. arXiv preprint arXiv:1606.04671 (2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  23. [23]

    Lifelong Learning with Dynamically Expandable Networks

    Yoon, J., Yang, E., Lee, J. & Hwang, S. J. Lifelong learning with dynamically expandable networks. arXiv preprint arXiv:1708.01547 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  24. [24]

    Fernando, C. et al. Pathnet: Evolution channels gradient descent in super neural networks. arXiv preprint arXiv:1701.08734 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  25. [25]

    M., Sluiter, M

    Dekhovich, A., Tax, D. M., Sluiter, M. H. & Bessa, M. A. Continual prune-and- select: class-incremental learning with specialized subnetworks. Applied Intelli- gence 1–16 (2023)

    work page 2023

  26. [26]

    Continual Learning via Neural Pruning

    Golkar, S., Kagan, M. & Cho, K. Continual learning via neural pruning. arXiv preprint arXiv:1903.04476 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 1903

  27. [27]

    & Zhu, Z

    Xu, J. & Zhu, Z. Reinforced continual learning. Advances in Neural Information Processing Systems 31 (2018)

    work page 2018

  28. [28]

    & Zhang, F

    Gao, Q., Luo, Z., Klabjan, D. & Zhang, F. Efficient architecture search for con- tinual learning. IEEE Transactions on Neural Networks and Learning Systems (2022)

    work page 2022

  29. [29]

    Networks of spiking neurons: The third generation of neural network models

    Maass, W. Networks of spiking neurons: The third generation of neural network models. Neural networks 10, 1659–1671 (1997)

    work page 1997

  30. [30]

    & Zhao, Z

    Zhao, F., Zeng, Y., Han, B., Fang, H. & Zhao, Z. Nature-inspired self-organizing collision avoidance for drone swarm based on reward-modulated spiking neural network. Patterns 3, 100611 (2022)

    work page 2022

  31. [31]

    Wu, Y. et al. Direct training for spiking neural networks: Faster, larger, better. In Proceedings of the AAAI conference on artificial intelligence , vol. 33, 1311– 1318 (2019)

    work page 2019

  32. [32]

    & Shi, L

    Wu, Y., Deng, L., Li, G., Zhu, J. & Shi, L. Spatio-temporal backpropagation for training high-performance spiking neural networks. Frontiers in neuroscience 12, 331 (2018)

    work page 2018

  33. [33]

    J., Siegelmann, H

    Patel, D., Hazan, H., Saunders, D. J., Siegelmann, H. T. & Kozma, R. Improved robustness of reinforcement learning policies upon conversion to spiking neuronal network platforms applied to atari breakout game. Neural Networks 120, 108– 115 (2019)

    work page 2019

  34. [34]

    & Zhao, Z

    Sun, Y., Zeng, Y., Zhao, F. & Zhao, Z. Multi-compartment neuron and popula- tion encoding improved spiking neural network for deep distributional reinforce- ment learning. arXiv preprint arXiv:2301.07275 (2023). 19

    work page arXiv 2023

  35. [35]

    & Sun, Y

    Zhao, Z., Zhao, F., Zhao, Y., Zeng, Y. & Sun, Y. A brain-inspired theory of mind spiking neural network improves multi-agent cooperation and competition. Patterns (2023)

    work page 2023

  36. [36]

    Zhao, R. et al. A framework for the general design and computation of hybrid neural networks. Nature communications 13, 1–12 (2022)

    work page 2022

  37. [37]

    M., Ramanathan, S

    Panda, P., Allred, J. M., Ramanathan, S. & Roy, K. Asp: Learning to forget with adaptive synaptic plasticity in spiking neural networks. IEEE Journal on Emerging and Selected Topics in Circuits and Systems 8, 51–64 (2017)

    work page 2017

  38. [38]

    & Shen, G

    Han, B., Zhao, F., Zeng, Y., Pan, W. & Shen, G. Enhancing efficient continual learning with dynamic structure development of spiking neural networks. arXiv preprint arXiv:2308.04749 (2023)

    work page arXiv 2023

  39. [39]

    & Jaffard, R

    Bontempi, B., Laurent-Demir, C., Destrade, C. & Jaffard, R. Time-dependent reorganization of brain circuitry underlying long-term memory storage. Nature 400, 671–675 (1999)

    work page 1999

  40. [40]

    & Xiang, T

    Zhou, K., Yang, Y., Hospedales, T. & Xiang, T. Deep domain-adversarial image generation for domain generalisation. In Proceedings of the AAAI conference on artificial intelligence, vol. 34, 13025–13032 (2020)

    work page 2020

  41. [41]

    & Hoi, S

    Pham, Q., Liu, C. & Hoi, S. Dualnet: Continual learning, fast and slow. Advances in Neural Information Processing Systems 34, 16131–16144 (2021)

    work page 2021

  42. [42]

    & Zhang, B.-T

    Lee, S.-W., Kim, J.-H., Jun, J., Ha, J.-W. & Zhang, B.-T. Overcoming catas- trophic forgetting by incremental moment matching. Advances in neural infor- mation processing systems 30 (2017)

    work page 2017

  43. [43]

    & Karatzoglou, A

    Serra, J., Suris, D., Miron, M. & Karatzoglou, A. Overcoming catastrophic for- getting with hard attention to the task. In International conference on machine learning, 4548–4557 (PMLR, 2018)

    work page 2018

  44. [44]

    & Xiong, C

    Li, X., Zhou, Y., Wu, T., Socher, R. & Xiong, C. Learn to grow: A continual structure learning framework for overcoming catastrophic forgetting. In Inter- national Conference on Machine Learning , 3925–3934 (PMLR, 2019)

    work page 2019

  45. [45]

    Wortsman, M. et al. Supermasks in superposition. Advances in Neural Informa- tion Processing Systems 33, 15173–15184 (2020)

    work page 2020

  46. [46]

    E., Hayasaka, S

    Joyce, K. E., Hayasaka, S. & Laurienti, P. J. The human functional brain net- work demonstrates structural and dynamical resilience to targeted attack. PLoS computational biology 9, e1002885 (2013)

    work page 2013

  47. [47]

    J., De Medinaceli, L

    Freed, W. J., De Medinaceli, L. & Wyatt, R. J. Promoting functional plasticity in the damaged nervous system. Science 227, 1544–1552 (1985). 20

    work page 1985

  48. [48]

    & Rosenfeld, J

    Anderson, V., Catroppa, C., Morse, S., Haritou, F. & Rosenfeld, J. Functional plasticity or vulnerability after early brain injury? Pediatrics 116, 1374–1382 (2005)

    work page 2005

  49. [49]

    residual vision activation theory

    Sabel, B. A., Henrich-Noack, P., Fedorov, A. & Gall, C. Vision restoration after brain and retina damage: the “residual vision activation theory”. Progress in brain research 192, 199–262 (2011)

    work page 2011

  50. [50]

    M., Peterson, B

    Xerri, C., Merzenich, M. M., Peterson, B. E. & Jenkins, W. Plasticity of primary somatosensory cortex paralleling sensorimotor skill recovery from stroke in adult monkeys. Journal of neurophysiology 79, 2119–2148 (1998)

    work page 1998

  51. [51]

    D., Elzinga, B., Schmahl, C

    Bremner, J. D., Elzinga, B., Schmahl, C. & Vermetten, E. Structural and func- tional plasticity of the human brain in posttraumatic stress disorder. Progress in brain research 167, 171–186 (2007)

    work page 2007

  52. [52]

    S., Duncan, L

    Guy, B., Zhang, J. S., Duncan, L. H. & Johnston Jr, R. J. Human neural organoids: Models for developmental neurobiology and disease. Developmental biology 478, 102–121 (2021)

    work page 2021

  53. [53]

    Zhang, L. I. & Poo, M.-m. Electrical activity and development of neural circuits. Nature neuroscience 4, 1207–1214 (2001)

    work page 2001

  54. [54]

    Arain, M. et al. Maturation of the adolescent brain. Neuropsychiatric disease and treatment 449–461 (2013)

    work page 2013

  55. [55]

    Huttenlocher, P. R. et al. Synaptic density in human frontal cortex- developmental changes and effects of aging. Brain Res 163, 195–205 (1979)

    work page 1979

  56. [56]

    Hartwigsen, G. et al. Rapid short-term reorganization in the language network. Elife 6, e25964 (2017)

    work page 2017

  57. [57]

    & Kirino, Y

    Takehara, K., Kawahara, S. & Kirino, Y. Time-dependent reorganization of the brain components underlying memory retention in trace eyeblink conditioning. Journal of Neuroscience 23, 9897–9905 (2003)

    work page 2003

  58. [58]

    & Shin, J

    Cha, H., Lee, J. & Shin, J. Co2l: Contrastive continual learning. InProceedings of the IEEE/CVF International conference on computer vision , 9516–9525 (2021)

    work page 2021

  59. [59]

    A., Pilly, P

    Ben-Iwhiwhu, E., Dick, J., Ketz, N. A., Pilly, P. K. & Soltoggio, A. Context meta-reinforcement learning via neuromodulation. Neural Networks 152, 70–79 (2022)

    work page 2022

  60. [60]

    Yan, S., Xie, J. & He, X. Der: Dynamically expandable representation for class incremental learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 3014–3023 (2021)

    work page 2021

  61. [61]

    Hung, C.-Y. et al. Compacting, picking and growing for unforgetting continual learning. Advances in Neural Information Processing Systems 32 (2019). 21

    work page 2019

  62. [62]

    Siddiqui, Z. A. & Park, U. Progressive convolutional neural network for incre- mental learning. Electronics 10, 1879 (2021)

    work page 2021

  63. [63]

    Abbott, L. F. Lapicque’s introduction of the integrate-and-fire model neuron (1907). Brain Research Bulletin 50, 303–304 (1999)

    work page 1907

  64. [64]

    Qin, H. et al. Forward and backward information retention for accurate binary neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2250–2259 (2020)

    work page 2020

  65. [65]

    DARTS: Differentiable Architecture Search

    Liu, H., Simonyan, K. & Yang, Y. Darts: Differentiable architecture search. arXiv preprint arXiv:1806.09055 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  66. [66]

    & Liu, T

    Li, Q., Wu, X. & Liu, T. Differentiable neural architecture search for optimal spatial/temporal brain function network decomposition. Medical Image Analysis 69, 101974 (2021)

    work page 2021

  67. [67]

    Xu, B., Wang, N., Chen, T. & Li, M. Empirical evaluation of rectified activations in convolutional network. arXiv preprint arXiv:1505.00853 (2015)

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  68. [68]

    Deng, J. et al. Imagenet: A large-scale hierarchical image database. In2009 IEEE conference on computer vision and pattern recognition , 248–255 (Ieee, 2009). Contributions B.H.,F.Z. and Y.Z. designed the study. B.H.,F.Z. W.P., Z.Z. and X.L.performed the experiments. B.H., F.Z., Z.Z.and X.L. counted and analyzed the experiment results. B.H.,F.Z., Q.K. and...

    work page 2009