Pith. sign in

REVIEW 5 major objections 5 minor 35 references

MTSpark: Enabling Multi-Task Learning with Spiking Neural Networks for Generalist Agents

T0 review · 5 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read MTSpark enables multi-task learning in spiking Q-networks by gating neurons with task-specific context signals.

desk verdict A useful combination of active dendrites and dueling in a spiking Q-network, but the catastrophic-forgetting claim is untested because the algorithm jointly trains on all tasks at every timestep. read the letter →

arxiv 2412.04847 v1 pith:SBF4KH4B submitted 2024-12-06 cs.NE cs.AIcs.LG

Avaneesh Devkota , Rachmad Vidya Wicaksana Putra , Muhammad Shafique This is my paper · ORCID
classification cs.NEcs.AIcs.LG
keywords spikingneuralnetworksmulti-tasklearningreinforcementcatastrophicforgettingactivedendritesdeepQ-networkduelingarchitecturecontinual
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

This paper proposes MTSpark, a method for training a single spiking neural network on multiple reinforcement-learning tasks without forgetting earlier ones. The key idea is to give each spiking neuron active dendrites that read a one-hot task-identity signal, so the network routes each task through its own sub-network while sharing most parameters. On three Atari games (Pong, Breakout, Enduro) the method reports scores close to or above human levels and clearly higher than DQN and DSQN baselines, and it also beats a conventional DNN on MNIST, Fashion-MNIST, and CIFAR-10 classification. The authors claim this is the first demonstration of multi-task learning in spiking neural networks under a reinforcement-learning setting, a step toward energy-efficient generalist agents.

What carries the argument

The central mechanism is the integrate-and-fire neuron with active dendrites: each dendritic segment $j$ has weights $d_j$ and reads the context signal $c$ (one-hot task identity), and the membrane potential updates as $V(t) = V(t-\Delta t) + f(\sum_i s_i(t), \max_j(d_j^T c))$, with $f(\sum_i s_i, \max_j(d_j^T c)) = \sum_i s_i \cdot \sigma(\max_j(d_j^T c))$. The max over dendrites makes each neuron selectively sensitive to the task identity, gating incoming spikes and forming per-task sub-networks. On top of this, the dueling structure in the MTSpark ADD variant computes a state value and per-action advantages separately before combining them into Q-values, which the paper argues improves action evaluation across tasks.

What would settle it

Train MTSpark with a strictly sequential protocol in which the inner loop over environments (lines 14-18 of Algorithm 1) samples only the current task's replay buffer, and compare Pong, Breakout, and Enduro scores with those reported; if the scores fall to baseline levels, the reported multi-task performance is explained by joint replay, not by active-dendrite gating.

Watch

Extended reading notes

Core claim

The paper's central discovery, stated on its own terms, is that context-signal-driven active dendrites let a deep spiking Q-network hold several tasks in one parameter set: when the one-hot task vector $c$ is fed into dendritic segments, the modulating function $f(\sum_i s_i(t), \max_j(d_j^T c))$ gates presynaptic spikes before they reach the membrane potential, so the network develops task-specific spiking patterns. Adding a dueling structure (separate state-value and advantage streams) on top of this gives the MTSpark ADD variant, which the paper reports achieves $-5.4$ in Pong, $0.6$ in Breakout, and $371.2$ in Enduro, outperforming DQN, DSQN, and their dueling variants. The same architecture also reaches 97.5% (MNIST), 86.4% (Fashion-MNIST), and 56% (CIFAR-10) accuracy, above the DNN comparison. The authors take these results as evidence that spiking networks can be made into generalist agents without growing the network.

Load-bearing premise

The claim that MTSpark prevents catastrophic forgetting assumes the benchmark actually presents tasks sequentially, but the training loop samples from all tasks' replay buffers at every timestep (Algorithm 1, lines 14-18), so the network trains on all tasks jointly and forgetting is never directly measured.

Editorial extensions

If this is right

  • A single spiking network can serve multiple RL tasks without adding task-specific parameters, since the one-hot context signal alone routes behavior.
  • The parameter counts of MTSpark AD (1,693,691) and MTSpark ADD (3,300,357) are nearly identical to their non-dendritic counterparts, so the multi-task capability does not come from added memory.
  • If the reported scores hold, spiking generalist agents become plausible for edge and robotic settings where energy-efficient multi-task adaptation is needed.
  • The dueling structure can be added on top of other spiking RL architectures to improve multi-task performance, as the comparison between MTSpark AD and MTSpark ADD suggests.
  • The same context-gating principle transfers to supervised image classification, indicating the mechanism is not specific to reinforcement learning.

Reading between the lines

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

  • Ablating the context signal (feeding the same vector for all tasks) would test how much of the reported performance depends on task identity gating rather than on the network's shared feature extractor.
  • Because the reported training loop samples from every task's replay buffer at each timestep, a strictly sequential version of the same algorithm would separate the contribution of active dendrites from the contribution of joint rehearsal.
  • The one-hot context signal presumes the task identity is known in advance; replacing it with a learned embedding or an inferred task descriptor would be a natural step toward task-agnostic multi-task RL.
  • The results cover three Atari games and three image sets; extending the evaluation to continuous-control or sparse-reward environments would map the limits of the gating approach.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. This paper proposes MTSpark, a spiking Q-network architecture that combines active dendrites with a dueling structure for multi-task reinforcement learning. The authors claim that task-specific context signals modulate dendritic segments, forming specialized sub-networks that can learn multiple Atari games without catastrophic forgetting, and also report accuracy improvements over DNN baselines on MNIST, Fashion-MNIST, and CIFAR-10. The evaluation compares MTSpark variants against DQN, DSQN, and their dueling versions over 250 episodes on Pong, Breakout, and Enduro, and against a DNN on classification tasks, along with ablations on two- and three-task classification scenarios.

Significance. If the results were reliable, showing that a single spiking network with active dendrites and dueling can learn multiple Atari tasks at human level would be a valuable step for energy-efficient generalist agents, particularly because the parameter overhead is negligible (Table 5). The paper also provides a clear algorithmic description and detailed architecture configurations. However, the central claim about catastrophic forgetting is not tested by the proposed training protocol, and several reported results are internally inconsistent or overstated. The significance of the contribution is therefore not established by the current evidence.

major comments (5)
  1. [Section 3.3, Algorithm 1] Algorithm 1 (lines 14-18) samples and updates from every environment's replay buffer at each timestep, so the network is trained jointly on all tasks; the text in Section 3.3 saying the network is trained on each environment for P episodes before switching is inconsistent with the algorithm. Consequently, the experiments never test catastrophic forgetting, which is the paper's central motivation. The comparison with DQN and DSQN is also confounded because their training protocol is not reported; if they were trained sequentially, the performance gap could be explained by replay alone rather than by active dendrites.
  2. [Abstract and Table 4] The abstract states that MTSpark reaches human-level performance on Breakout with a score of 0.6 against the human score of 31, but Table 4 and Section 5.1 show that no model, including MTSpark ADD, approaches the human score in Breakout (the best reported score is DQN at 3.2). This is a direct misrepresentation of the reported results and overstates the central experimental claim.
  3. [Section 1.3 vs Section 5.2] The Key Results in the Introduction report the state-of-the-art MNIST accuracy as 33.9%, while Section 5.2 and Figure 8 report the DNN baseline reaching approximately 57% accuracy on MNIST. No explanation reconciles these numbers, so the claimed superior accuracy of MTSpark over the baseline is ambiguous and the abstract's accuracy comparisons are not reproducible from the text.
  4. [Section 5 (all experiments)] All RL and classification results are reported as single runs without error bars, confidence intervals, or multiple seeds. Given the known high variance of Atari RL and the small gaps between some models (e.g., Enduro 371.2 vs 362.2, Pong -5.4 vs -9.4), the statistical significance of the claimed improvements is not established, and the results cannot be considered reliable evidence for the superiority of MTSpark.
  5. [Section 5.3.3] The ablation study compares MTSpark, which uses spiking neurons with active dendrites, against a DNN with ReLU activations, without a spiking no-dendrite control and without specifying whether training was sequential or joint. This confounds the effect of active dendrites with architecture differences and replay strategy, so the mechanism claimed for the improvements is not isolated.
minor comments (5)
  1. [Table 3] The replay buffer size is listed as '220' in Table 3 and in the text; this should be clarified as 2^20 or the correct numeric value.
  2. [Table 2] In the advantage branch, the row 'Non-Spiking Neuron 512 ×1 18 ×1' appears to be a typo; the input dimension should be 18×1, not 512×1.
  3. [Section 3.1] There is a typo in the sentence preceding Equation 1: 'whise behavior' should be 'whose behavior'.
  4. [Section 5.3.3] There are typos in this subsection: 'convetional' should be 'conventional' and 'Fahion' should be 'Fashion'.
  5. [Section 3.1 and Table 3] The membrane threshold V_th and the number of simulation timesteps are not specified anywhere, which are essential hyperparameters for reproducing spiking network results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MTSpark's contributions are empirical and its design choices are not fitted to the reported outcomes.

full rationale

This paper proposes an SNN architecture (MTSpark) with active dendrites and a dueling structure, and evaluates it on Atari games and image classification. There is no derivation chain whose conclusion is equivalent to its premises. The modulating function in Equation 3 is a defined architectural choice, not a quantity fitted to the reported scores, and the context signals are fixed one-hot vectors. The claimed performance gains are presented as experimental results against DQN, DSQN, and dueling variants, so the central claim is externally benchmarked rather than self-referential. The active-dendrite mechanism is motivated by prior independent work (Iyer et al. and Pes et al.), and the authors' own prior SNN papers are cited only in related-work context and are not load-bearing for the main result. The training procedure in Algorithm 1 updates the network on all environments' replay buffers every timestep, which raises a genuine methodological concern about whether catastrophic forgetting is actually tested, but that is an experimental-validity issue, not a circularity issue: the result is not forced by construction, and no fitted parameter is renamed as a prediction. Accordingly, no circular step can be exhibited, and the appropriate finding is no significant circularity.

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

The paper introduces no new entities. Its central claims rely on design choices (dendrite count, modulation function, training schedule) that are stated but not derived, and on the assumption that the joint training protocol is a valid test of catastrophic forgetting.

free parameters (4)
  • Number of active dendrites per neuron = 3 (equal to number of tasks)
    Chosen for efficient implementation, not derived from data.
  • Modulating function = sigmoid of max dendritic activation
    Ad hoc choice; no comparison to alternative gating functions.
  • Membrane threshold Vth = Not specified in the paper
    Critical SNN hyperparameter omitted from the text.
  • Number of simulation timesteps = Not specified
    The paper does not state how many timesteps the spiking network is simulated for, which is essential for SNN reproducibility.
assumptions (4)
  • domain assumption Active dendrites as defined in Eq. 1-3 improve multi-task learning
    The paper assumes this mechanism produces task-specific sub-networks; no theoretical guarantee is given.
  • domain assumption Dueling structure improves performance in SNN context
    Borrowed from DNN dueling networks [34]; assumed to transfer to spiking Q-networks.
  • ad hoc to paper The joint replay training strategy in Algorithm 1 is an appropriate test for multi-task learning
    The paper frames the problem as catastrophic forgetting but uses a training scheme that continuously updates on all tasks, which avoids forgetting by construction.
  • domain assumption Task identity (context signal) is available at inference time
    The one-hot context vector is provided as input; a generalist agent may not always know which task it faces.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MTSpark: Enabling Multi-Task Learning with Spiking Neural Networks for Generalist Agents." pith.science (2026). https://pith.science/paper/SBF4KH4B

@misc{pith2026241204847,
  author       = {Pith},
  title        = {Pith review of: MTSpark: Enabling Multi-Task Learning with Spiking Neural Networks for Generalist Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SBF4KH4B}},
  note         = {Machine review of arXiv:2412.04847}
}
read the original abstract

Currently, state-of-the-art RL methods excel in single-task settings, but they still struggle to generalize across multiple tasks due to catastrophic forgetting challenges, where previously learned tasks are forgotten as new tasks are introduced. This multi-task learning capability is significantly important for generalist agents, where adaptation features are highly required (e.g., autonomous robots). On the other hand, Spiking Neural Networks (SNNs) have emerged as alternative energy-efficient neural network algorithms due to their sparse spike-based operations. Toward this, we propose MTSpark, a novel methodology to enable multi-task RL using spiking networks. Specifically, MTSpark develops a Deep Spiking Q-Network (DSQN) with active dendrites and dueling structure by leveraging task-specific context signals. Specifically, each neuron computes task-dependent activations that dynamically modulate inputs, forming specialized sub-networks for each task. Moreover, this bioplausible network model also benefits from SNNs, enhancing energy efficiency and making the model suitable for hardware implementation. Experimental results show that, our MTSpark effectively learns multiple tasks with higher performance compared to the state-of-the-art. Specifically, MTSpark successfully achieves high score in three Atari games (i.e., Pong: -5.4, Breakout: 0.6, and Enduro: 371.2), reaching human-level performance (i.e., Pong: -3, Breakout: 31, and Enduro: 368), where state-of-the-art struggle to achieve. In addition, our MTSpark also shows better accuracy in image classification tasks than the state-of-the-art. These results highlight the potential of our MTSpark methodology to develop generalist agents that can learn multiple tasks by leveraging both RL and SNN concepts.

Figures

Figures reproduced from arXiv: 2412.04847 by the authors.

Figure 1
Figure 1. Multi-task learning performance of the state-of-the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overview of our MTSpark methodology, showing three key steps: spiking neuron development with active dendrites, network [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The overview of a DQN architecture. multi-task learning in real-world scenarios [12, 14, 15]. To￾ward this, many methods have been developed [15, 33], and the state-of-the-art include the architectural-, replay-, and regularization-based methods. 2.2.1. Architectural-based Methods These methods employ techniques where a dedicated archi￾tecture enhancement (e.g., layer) is added for each new task learning, hence redu… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The proposed integrate-and-fire (IF) neuron model is [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 6
Figure 6. Figure 6: The configuration of our proposed architecture variants: [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Performance of different models (i.e., DQN, DSQN, DQN [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Test accuracy of different network models: (a) DNN and [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Test accuracy in the two-class learning scenario for [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Test accuracy in the three-class learning scenario [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 22 canonical work pages

  1. [1]

    Controlled forgetting: Tar- geted stimulation and dopaminergic plasticity modulation for unsupervised lifelong learning in spiking neural networks

    Jason M Allred and Kaushik Roy. Controlled forgetting: Tar- geted stimulation and dopaminergic plasticity modulation for unsupervised lifelong learning in spiking neural networks. Frontiers in neuroscience, 14:7, 2020. 2

    work page 2020

  2. [2]

    Em- bodied neuromorphic intelligence

    Chiara Bartolozzi, Giacomo Indiveri, and Elisa Donati. Em- bodied neuromorphic intelligence. Nature communications, 13(1):1024, 2022. 2

    work page 2022

  3. [3]

    Deep reinforcement learning with spiking q-learning

    Ding Chen et al. Deep reinforcement learning with spiking q-learning. CoRR, abs/2201.09754, 2022. 1, 2, 6

    arXiv 2022

  4. [4]

    Ob- ject detection with spiking neural networks on automotive event data

    Loic Cordone, Beno ˆıt Miramond, and Phillipe Thierion. Ob- ject detection with spiking neural networks on automotive event data. In International Joint Conference on Neural Net- works (IJCNN), pages 1–8, 2022. 2

    work page 2022

  5. [5]

    The mnist database of handwritten digit images for machine learning research [best of the web]

    Li Deng. The mnist database of handwritten digit images for machine learning research [best of the web]. IEEE Signal Processing Magazine, 29(6):141–142, 2012. 7

    work page 2012

  6. [6]

    Avoiding catastrophe: Active dendrites enable multi-task learning in dynamic environments

    Abhiram Iyer et al. Avoiding catastrophe: Active dendrites enable multi-task learning in dynamic environments. Fron- tiers in Neurorobotics, 16, 2022. 3

    work page 2022

  7. [7]

    Overcoming catastrophic forgetting in neural networks

    James Kirkpatrick et al. Overcoming catastrophic forgetting in neural networks. CoRR, abs/1612.00796, 2016. 3

    arXiv 2016

  8. [8]

    The cifar-10 dataset

    Alex Krizhevsky, Vinod Nair, Geoffrey Hinton, et al. The cifar-10 dataset. online: http://www. cs. toronto. edu/kriz/cifar. html, 55(5):2, 2014. 7

    work page 2014

Show all 35 references
  1. [9]

    Lifelong learning with dynamically ex- pandable networks

    Jeongtae Lee et al. Lifelong learning with dynamically ex- pandable networks. CoRR, abs/1708.01547, 2017. 2

    2017 arXiv

  2. [10]

    Gradi- ent episodic memory for continuum learning

    David Lopez-Paz and Marc’Aurelio Ranzato. Gradi- ent episodic memory for continuum learning. CoRR, abs/1706.08840, 2017. 2

    2017 arXiv

  3. [11]

    Eeg-based emotion classification using spiking neural networks

    Yuling Luo et al. Eeg-based emotion classification using spiking neural networks. IEEE Access , 8:46007–46016,

  4. [12]

    Online continual learning in image classi- fication: An empirical survey

    Zheda Mai et al. Online continual learning in image classi- fication: An empirical survey. Neurocomputing, 469:28–51,

  5. [13]

    Michael McCloskey and Neal J. Cohen. Catastrophic inter- ference in connectionist networks: The sequential learning problem. pages 109–165. Academic Press, 1989. 1, 3

    1989

  6. [14]

    The stability-plasticity dilemma: Investigating the contin- uum from catastrophic forgetting to age-limited learning ef- fects

    Martial Mermillod, Aur ´elia Bugaiska, and Patrick Bonin. The stability-plasticity dilemma: Investigating the contin- uum from catastrophic forgetting to age-limited learning ef- fects. Frontiers in psychology, 4:54654, 2013. 3

    2013

  7. [15]

    Continual learning with neu- romorphic computing: Theories, methods, and applications

    Mishal Fatima Minhas et al. Continual learning with neu- romorphic computing: Theories, methods, and applications. arXiv preprint arXiv:2410.09218, 2024. 1, 2, 3

    2024 arXiv

  8. [16]

    Playing atari with deep reinforcement learning

    V olodymyr Mnih et al. Playing atari with deep reinforcement learning. CoRR, abs/1312.5602, 2013. 1, 2, 3, 5, 6, 7

    2013 arXiv

  9. [17]

    Human-level control through deep reinforcement learning

    V olodymyr Mnih et al. Human-level control through deep reinforcement learning. Nature, 518(7540):529–533, 2015. 1, 5

    2015

  10. [18]

    Active dendrites enable efficient continual learning in time-to-first-spike neural networks, 2024

    Lorenzo Pes et al. Active dendrites enable efficient continual learning in time-to-first-spike neural networks, 2024. 2, 3

    2024

  11. [19]

    Rachmad Vidya Wicaksana Putra and Muhammad Shafique. FSpiNN: An optimization framework for memory-efficient and energy-efficient spiking neural networks.IEEE Transac- tions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 39(11):3601–3613, 2020

    2020

  12. [20]

    Q-spinn: A framework for quantizing spiking neural net- works

    Rachmad Vidya Wicaksana Putra and Muhammad Shafique. Q-spinn: A framework for quantizing spiking neural net- works. In 2021 International Joint Conference on Neural Networks (IJCNN), pages 1–8, 2021. 2

    2021

  13. [21]

    Spikedyn: A framework for energy-efficient spiking neural networks with continual and unsupervised learning capabil- ities in dynamic environments

    Rachmad Vidya Wicaksana Putra and Muhammad Shafique. Spikedyn: A framework for energy-efficient spiking neural networks with continual and unsupervised learning capabil- ities in dynamic environments. In 2021 58th ACM/IEEE Design Automation Conference (DAC) , pages 1057–1062. ...

    2021

  14. [22]

    lpspikecon: Enabling low-precision spiking neural network processing for efficient unsupervised continual learning on autonomous agents

    Rachmad Vidya Wicaksana Putra and Muhammad Shafique. lpspikecon: Enabling low-precision spiking neural network processing for efficient unsupervised continual learning on autonomous agents. In 2022 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE, 2022. 2

    2022

  15. [23]

    Snn4agents: A framework for devel- oping energy-efficient embodied spiking neural networks for autonomous agents

    Rachmad Vidya Wicaksana Putra, Alberto Marchisio, and Muhammad Shafique. Snn4agents: A framework for devel- oping energy-efficient embodied spiking neural networks for autonomous agents. Frontiers in Robotics and AI (FROBT), 11, 2024. 2

    2024

  16. [24]

    Embodied neu- romorphic artificial intelligence for robotics: Perspectives, challenges, and research development stack

    Rachmad Vidya Wicaksana Putra et al. Embodied neu- romorphic artificial intelligence for robotics: Perspectives, challenges, and research development stack. arXiv preprint arXiv:2404.03325, 2024. 2

    2024 arXiv

  17. [25]

    A generalist agent, 2022

    Scott Reed et al. A generalist agent, 2022. 2

    2022

  18. [26]

    Learning to learn without forgetting by maximizing transfer and minimizing interference

    Matthew Riemer et al. Learning to learn without forgetting by maximizing transfer and minimizing interference. CoRR, abs/1810.11910, 2018. 2

    2018 arXiv

  19. [27]

    Overcoming catastrophic forgetting with hard attention to the task

    Joan Serr `a et al. Overcoming catastrophic forgetting with hard attention to the task. CoRR, abs/1801.01423, 2018. 2

    2018 arXiv

  20. [28]

    Continual learning with deep generative replay

    Hanul Shin et al. Continual learning with deep generative replay. CoRR, abs/1705.08690, 2017. 3

    2017 arXiv

  21. [29]

    Mastering chess and shogi by self-play with a general reinforcement learning algorithm

    David Silver et al. Mastering chess and shogi by self-play with a general reinforcement learning algorithm. CoRR, abs/1712.01815, 2017. 1

    2017 arXiv

  22. [30]

    Multi-task reinforcement learning with context-based representations

    Shagun Sodhani, Amy Zhang, and Joelle Pineau. Multi-task reinforcement learning with context-based representations. CoRR, abs/2102.06177, 2021. 1

    2021 arXiv

  23. [31]

    Distral: Robust multitask reinforcement learning

    Yee Whye Teh et al. Distral: Robust multitask reinforcement learning. CoRR, abs/1707.04175, 2017. 2

    2017 arXiv

  24. [32]

    Nelson Vithayathil Varghese and Qusay H. Mahmoud. A sur- vey of multi-task deep reinforcement learning. Electronics, 9(9), 2020. 2

    2020

  25. [33]

    A comprehensive survey of continual learning: Theory, method and application

    Liyuan Wang et al. A comprehensive survey of continual learning: Theory, method and application. IEEE Transac- tions on Pattern Analysis and Machine Intelligence, 2024. 1, 2, 3

    2024

  26. [34]

    Dueling network architectures for deep reinforcement learning

    Ziyu Wang et al. Dueling network architectures for deep reinforcement learning. In International Conference on Ma- chine Learning (ICML), pages 1995–2003. PMLR, 2016. 5

    1995

  27. [35]

    Fashion- mnist: a novel image dataset for benchmarking machine learning algorithms

    Han Xiao, Kashif Rasul, and Roland V ollgraf. Fashion- mnist: a novel image dataset for benchmarking machine learning algorithms. CoRR, abs/1708.07747, 2017. 7

    2017 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.