Pith. sign in

REVIEW 4 major objections 5 minor 57 references

Inherently Interpretable and Uncertainty-Aware Models for Online Learning in Cyber-Security Problems

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper argues that Additive Gaussian Processes, retrained on a rolling buffer of recent and randomly drawn historical data, can scale to online cyber-security classification while keeping interpretability and uncertainty…

desk verdict A competent application of AGPs to online phishing classification whose feasibility claim survives but whose advertised scalability fix is never actually measured. read the letter →

arxiv 2411.09393 v1 pith:6KOHJK7L submitted 2024-11-14 cs.LG

Benjamin Kolicic , Alberto Caron , Chris Hicks , Vasilios Mavroudis This is my paper · ORCID
classification cs.LG
keywords additiveGaussianprocessesinterpretablemachinelearninguncertaintyquantificationonlinecyber-securityphishingdetectionactivegeneralizedmodels
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

The paper tries to establish that Additive Gaussian Processes (AGPs)—Gaussian-process versions of generalized additive models, where each feature gets its own kernel and its own function—can be made practical for online cyber-security classification. Its proposed fix is a rolling buffer: at each update, retrain on the most recent batch plus randomly sampled historical batches, instead of the whole dataset. On a URL phishing classification task with over 11,000 samples and 30 features, the AGP reaches AUC 0.952 and F1 scores in the same range as the other models, close to a black-box neural network (AUC 0.978), while providing per-feature contributions and posterior variances the network cannot. If the claim holds, security analysts get a model that is transparent by construction and uncertainty-aware without paying a large accuracy penalty, and the scalability objection that has blocked AGP adoption is eased.

What carries the argument

The load-bearing object is the additive kernel $k(x,x') = k_0 + \sum_{j=1}^p k_j(x_j,x_j')$, which makes a Gaussian Process take the generalized additive form $g(E[Y]) = \beta_0 + \sum_{j=1}^p f_j(x_j)$. Because each feature has its own independent kernel, the per-feature function $f_j(x_j)$ can be recovered and interpreted as a Shapley value, and it carries its own posterior variance, giving both interpretability and uncertainty quantification. The scalability mechanism is the rolling buffer: retraining on the most recent data plus randomly drawn historical data, with the recent-to-historical ratio a tunable hyperparameter, instead of retraining on the full dataset or learning sparse inducing points at every step.

What would settle it

Run the same pipeline on a timestamped stream of real URLs or network connections in chronological order, compare its F1 and AUC against an offline model retrained on all data, and measure the autocorrelation of labels and features. A material drop in performance, or a buffer that cannot track drift, would refute the claim.

Watch

Extended reading notes

Core claim

The central claim is that the rolling-buffer pipeline eases the scaling issue of AGPs "without incurring a significant loss in terms of model performance," in the paper's words. The demonstration is an online URL phishing classification problem: models are warmed up on an initial window, then updated with random batches of 50–100 new URLs, and optionally with uncertainty-based active learning that labels only the most uncertain samples. Across window proportions from roughly 10% to 100%, the AGP's F1 score stays close to those of a full GP, a Neural Additive Model, and a neural network, and at the reported 20% window its AUC is 0.952 versus 0.978 for the neural network. The AGP also concentrates its posterior variance on a small set of meaningful features—absence of HTTPS and absence of an anchor URL—while the neural additive model disperses its variance across many features, which the paper reads as evidence that NAM suffers from the over-confidence typical of deep networks. The paper positions AGP as the first inherently interpretable and uncertainty-aware model class applied to cyber-security online learning.

Load-bearing premise

Everything rests on the assumption that consecutive samples in the online stream are independent; real security traffic arrives in correlated bursts, so the reported numbers may not survive a chronological deployment.

Editorial extensions

If this is right

  • At a window proportion of 20–30%, models keep most of their full-data F1, so the buffer defines a practical operating point for online retraining where the computational saving is large and the accuracy cost is small.
  • AGP outputs can be read as per-feature Shapley contributions, so an analyst can see directly that missing HTTPS and missing anchor URLs drive phishing decisions, without running a post-hoc explainer.
  • The posterior variance gives a principled acquisition function for active learning: labelling only 10–20% of each new batch maintains performance, lowering labelling cost.
  • Because the GP posterior separates epistemic from aleatoric uncertainty, the model can flag out-of-sample inputs, something the neural baselines cannot do reliably.
  • The additive-kernel construction extends naturally to regression and to pairwise interaction terms, so the pipeline is not limited to binary phishing detection.

Reading between the lines

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

  • The rolling-buffer recipe is model-agnostic: any GP variant whose bottleneck is the $O(n^3)$ covariance inversion could be dropped into the same online loop, so the paper's scalability fix may carry over to sparse, deep, or multi-output GPs without new theory.
  • The independence assumption is likely the first thing to break in deployment: real security event streams are autocorrelated, and the reported metrics come from randomly drawn batches, so a chronological re-run on real traffic is the natural stress test.
  • The feature-specific variance signal suggests a cheap novelty detector: flag samples whose highest-variance feature shifts away from the features the model relies on, and route those to human analysts; this is a direct, testable use of the AGP output that the paper does not implement.
  • If the result transfers, the main practical competitor is not the neural network but the Neural Additive Model, since NAM matches the interpretability at lower computational cost; the decisive advantage claimed for AGP is exactly the uncertainty quantification that NAM lacks, so deployment decisions hinge on how much that uncertainty is worth in practice.
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

4 major / 5 minor

Summary. The paper proposes using Additive Gaussian Processes (AGPs) and Neural Additive Models (NAMs) for online supervised learning in cybersecurity, specifically phishing URL classification. The central claim is that a rolling buffer of recent plus randomly drawn historical data eases AGPs' cubic scaling without incurring significant performance loss (Abstract, §3.3), while retaining inherent interpretability and uncertainty quantification. The authors compare AGP, GP, NAM, and NN on a Kaggle phishing dataset (11,000+ URLs, 30 features) using ROC/AUC and F1 curves across varying window proportions and active-learning label budgets (§4). They also show feature-level contribution and variance plots (§4.4) to illustrate interpretability and uncertainty awareness.

Significance. If the central claim were fully supported, the paper would be a useful practical contribution: it would show that an inherently interpretable, uncertainty-aware model can be applied online to a cybersecurity task at competitive accuracy with a concrete mechanism (rolling buffer) to mitigate the main known drawback of full GPs. The paper explicitly provides code (Abstract), and the model family (AGPs) is standard and correctly cited to external work, so the methodological core is sound. However, the significance is currently limited by the evidence provided: the scalability claim is asserted without timing or complexity measurements, the performance claims lack error bars and statistical tests, and the active-learning experiment lacks a random baseline. The interpretability and uncertainty visualizations are interesting but qualitative. The paper contributes a plausible pipeline, yet the advertised advantages over standard GPs and over non-additive baselines are not rigorously demonstrated.

major comments (4)
  1. [§3.3, §4.3] The central scalability claim that the rolling buffer 'eases the scaling issue' of AGPs is never directly tested. The paper reports F1 versus window proportion (Fig. 6) and versus labelled batch proportion (Fig. 7), but no runtime, memory, or FLOPs measurements are presented, and there is no comparison against retraining on the full dataset. Since §2.1 states the O(n^3) cost of GPs, the reader cannot assess whether a 20% window (about 2,200 points) is actually manageable for online cyber-security workloads. This is a load-bearing omission: the abstract and conclusion both rest on the claimed computational improvement.
  2. [§4.2, Fig. 5] The statement that 'no largely statistically significant difference' exists between the models is unsupported: the AUC values (NN 0.978, GP 0.955, NAM 0.969, AGP 0.952) are single-run numbers with no error bars, confidence intervals, or statistical tests. The same issue affects Figs. 6 and 7, where F1 curves are shown without repeated seeds or variance estimates, so the claim of 'no significant losses' at small window proportions is not quantified.
  3. [§4.3, Fig. 7] The active-learning experiment does not include a random-labelling baseline. Without comparing uncertainty-sampling to randomly selecting labels from the new batch, the conclusion that 'uncertainty-based active sampling is able to select highly representative samples' is not established. The observed upward trend in F1 as the labelled proportion increases could equally occur with random selection.
  4. [§2, §4.1] The online-learning setup assumes consecutive samples are independent: '(xt, yt) is independent of (xt+1, yt+1)' (§2). However, the experiments explicitly draw 'a random batch of new URLs' (§4.1), which removes any temporal ordering present in the original dataset. Real security event streams are temporally correlated (e.g., campaign-style phishing attacks), so the reported F1/AUC and the rolling-buffer design may not transfer to actual online deployment. The manuscript provides no experiment with temporally ordered data, which weakens the claim of applicability to 'the dynamic learning environment' of cyber-security.
minor comments (5)
  1. [Eq. (1)] Equation (1) contains a typographical error: 'g(E[Y ]) = β0 + Pp j=1 fj(xj) = (1) = β0 + f1(x1) + ...' has a duplicated equals sign and an unnecessary '(1)=' fragment that should be removed.
  2. [§3] The statement that the input-specific functions fj(xj) can be interpreted 'as Shapley values' is imprecise. GAM components are additive feature contributions, but they are not the Shapley values of the underlying game in general; the connection requires additional assumptions. Either clarify the sense in which this holds or cite a source that establishes it.
  3. [§4.1] The experimental setup does not state how many runs are averaged for the F1 and AUC results, nor whether the same train/test splits are used across models. Reporting the number of independent repetitions and the variance of the metrics would substantially improve the paper.
  4. [Abstract] The abstract says 'The source code is available here' but the full text only says 'available here' without a visible URL. Please include an explicit link or repository identifier.
  5. [Fig. 5] The ROC curves in Fig. 5 are not labelled in the legend in a way that allows the reader to match each curve to the AUC values listed in the text; consider adding the AUC values directly to the legend.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the AGP formulation, GAM decomposition, and rolling-buffer experiments are not derived from the paper's own conclusions.

full rationale

The paper's central claims are (i) AGPs are inherently interpretable and uncertainty-aware, and (ii) a rolling buffer of recent plus randomly drawn historical data eases AGP scalability without significant performance loss. Neither claim is established by defining a quantity in terms of the target result. The AGP posterior and additive-kernel GAM decomposition in Sections 2.1 and 3.2 are standard results cited to external work (Duvenaud et al. 2011; Rasmussen & Williams 2006; Hastie 2017), and the per-feature contributions are model outputs interpreted as Shapley values, not fitted to the paper's conclusions. The rolling-buffer design in Section 3.3 is an empirical pipeline choice, not a parameter fitted to the F1 results, so Figures 6 and 7 compare independent model behaviours rather than restating an input. The author self-citations in Sections 1 and 5 are contextual and related-work references, and they do not carry the load of the scalability or performance arguments. The manuscript does contain evidential gaps: Section 4.2 states "no largely statistically significant difference" without reporting error bars or statistical tests, and Section 3.3 claims "computational feasibility" while no runtime or memory experiment is reported; Section 6 itself concedes that AGP computational complexity "remains a significant challenge". These are correctness and evidence concerns, not circular reductions, because no equation or fitted value is equivalent by construction to a reported prediction.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests mainly on standard GP/GAM machinery plus two unproven practical assumptions: sample independence in the online simulation and the rolling buffer's ability to preserve global structure. No new theoretical entities are introduced.

free parameters (4)
  • Window Proportion = 0.2 (default), varied in Figure 6
    Fraction of the training data used for retraining; chosen by hand and directly affects the F1 versus computation trade-off (Section 3.3, Section 4.3).
  • Rolling buffer composition ratio = not reported
    Ratio of recent batches to randomly drawn historical batches is described as a problem-specific hyperparameter in Section 3.3, but no value is reported.
  • GP kernel hyperparameters = not reported
    Signal variance and length-scale of the SE kernel are optimized via marginal likelihood (Section 2.1) but the fitted values are not given.
  • Network architectures and training hyperparameters for NN and NAM = not reported
    The neural baselines require architecture and optimizer choices that are not specified, making the comparison difficult to reproduce (Section 4.2).
assumptions (5)
  • domain assumption Consecutive samples in the online setting are independent
    Section 2 states '(xt,yt) is independent of (xt+1,yt+1)', which justifies the random-batch simulation but may not hold for real cyber-security streams.
  • standard math The additive kernel decomposition yields independent 1-D GP classifiers equivalent to the GAM form
    Section 3.2 relies on the standard AGP result from Duvenaud et al. [16]; the paper does not reprove it.
  • ad hoc to paper Rolling buffer with recent plus randomly drawn historical batches preserves global patterns
    Section 3.3 asserts this balance without proof or direct measurement; it is central to the scalability claim.
  • standard math Sigmoid squashing of the GP latent function gives a valid probabilistic classification
    Section 2.1 uses the standard GP classification approximation; its calibration is not evaluated.
  • domain assumption The Kaggle URL dataset labels and 30 features are reliable proxies for real phishing data
    Section 4.1 uses a single scraped dataset; label quality and feature completeness are not independently verified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Inherently Interpretable and Uncertainty-Aware Models for Online Learning in Cyber-Security Problems." pith.science (2026). https://pith.science/paper/6KOHJK7L

@misc{pith2026241109393,
  author       = {Pith},
  title        = {Pith review of: Inherently Interpretable and Uncertainty-Aware Models for Online Learning in Cyber-Security Problems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6KOHJK7L}},
  note         = {Machine review of arXiv:2411.09393}
}
read the original abstract

In this paper, we address the critical need for interpretable and uncertainty-aware machine learning models in the context of online learning for high-risk industries, particularly cyber-security. While deep learning and other complex models have demonstrated impressive predictive capabilities, their opacity and lack of uncertainty quantification present significant questions about their trustworthiness. We propose a novel pipeline for online supervised learning problems in cyber-security, that harnesses the inherent interpretability and uncertainty awareness of Additive Gaussian Processes (AGPs) models. Our approach aims to balance predictive performance with transparency while improving the scalability of AGPs, which represents their main drawback, potentially enabling security analysts to better validate threat detection, troubleshoot and reduce false positives, and generally make trustworthy, informed decisions. This work contributes to the growing field of interpretable AI by proposing a class of models that can be significantly beneficial for high-stake decision problems such as the ones typical of the cyber-security domain. The source code is available.

Figures

Figures reproduced from arXiv: 2411.09393 by the authors.

Figure 1
Figure 1. Simple example of Gaussian Process fit one-dimensional input space. The red line denotes the mean fit for [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Example of Gaussian Process fit on a two-dimensional inputs case. The 3D plot on the left depicts the mean [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Architecture of a NAM: each input is modelled via an input-specific fully-connected MLP (sub-networks) [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Architecture of an Additive GP: each data feature is fed into a Multidimensional GP, then individual [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: ROC curves of the four models considered: Neural Nets (NN), Gaussian Processes (GP), Neural Additive [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: F1 classification scores of the four models (NN, GP, NAM and AGP), for increasing window proportion [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: F1 classification scores of the four models (NN, GP, NAM and AGP), for an increasing proportion of labelled [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Percentage of times a certain feature x ∈ X from the dataset gets picked by the models as the main contributor to the final classification. Feature 7 and 13 are the most important ones picked by both models and correspond to “URL not being HTTPS" and “URL not being an …
Figure 9
Figure 9. Figure 9: Percentage of times a certain feature x ∈ X from the dataset gets picked by the models as the feature with the highest estimated variance of classification contribution. AGP selects feature 13, “URL not being an Anchor URL", as the most variable one most of the time, w…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 34 canonical work pages

  1. [1]

    A review of uncertainty quantification in deep learning: Techniques, applications and challenges

    Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul Fieguth, Xiaochun Cao, Abbas Khosravi, U Rajendra Acharya, et al. A review of uncertainty quantification in deep learning: Techniques, applications and challenges. Information fusion, 76:243–297, 2021

    2021

  2. [2]

    why should i trust your ids?

    Zakaria Abou El Houda, Bouziane Brik, and Lyes Khoukhi. “why should i trust your ids?”: An explainable deep learning framework for intrusion detection systems in internet of things networks. IEEE Open Journal of the Communications Society, 3:1164–1176, 2022. 11

    work page 2022

  3. [3]

    Neural additive models: Interpretable machine learning with neural nets

    Rishabh Agarwal, Levi Melnick, Nicholas Frosst, Xuezhou Zhang, Ben Lengerich, Rich Caruana, and Geoffrey E Hinton. Neural additive models: Interpretable machine learning with neural nets. Advances in neural information processing systems, 34:4699–4711, 2021

    2021

  4. [4]

    On the effective- ness of machine and deep learning for cyber security

    Giovanni Apruzzese, Michele Colajanni, Luca Ferretti, Alessandro Guido, and Mirco Marchetti. On the effective- ness of machine and deep learning for cyber security. In 2018 10th international conference on cyber Conflict (CyCon), pages 371–390. IEEE, 2018

    work page 2018

  5. [5]

    Explainable artificial intelligence (xai): Concepts, taxonomies, opportunities and challenges toward responsible ai

    Alejandro Barredo Arrieta, Natalia Díaz-Rodríguez, Javier Del Ser, Adrien Bennetot, Siham Tabik, Alberto Barbado, Salvador García, Sergio Gil-López, Daniel Molina, Richard Benjamins, et al. Explainable artificial intelligence (xai): Concepts, taxonomies, opportunities and challenges toward responsible ai. Information fusion, 58:82–115, 2020

    work page 2020

  6. [6]

    Reward shaping for happier autonomous cyber security agents

    Elizabeth Bates, Vasilios Mavroudis, and Chris Hicks. Reward shaping for happier autonomous cyber security agents. In Proceedings of the 16th ACM Workshop on Artificial Intelligence and Security, pages 221–232, 2023

    work page 2023

  7. [7]

    A survey of deep learning methods for cyber security

    Daniel S Berman, Anna L Buczak, Jeffrey S Chavis, and Cherita L Corbett. A survey of deep learning methods for cyber security. Information, 10(4):122, 2019

    work page 2019

  8. [8]

    Machine learning explainability in finance: an application to default risk analysis

    Philippe Bracke, Anupam Datta, Carsten Jung, and Shayak Sen. Machine learning explainability in finance: an application to default risk analysis. Bank of England Working Paper, 2019

    work page 2019

Show all 57 references
  1. [9]

    The malicious use of artificial intelligence: Forecasting, prevention, and mitigation

    Miles Brundage, Shahar Avin, Jack Clark, Helen Toner, Peter Eckersley, Ben Garfinkel, Allan Dafoe, Paul Scharre, Thomas Zeitzoff, Bobby Filar, et al. The malicious use of artificial intelligence: Forecasting, prevention, and mitigation. arXiv preprint arXiv:1802.07228, 2018

    2018 arXiv

  2. [10]

    A survey of data mining and machine learning methods for cyber security intrusion detection

    Anna L Buczak and Erhan Guven. A survey of data mining and machine learning methods for cyber security intrusion detection. IEEE Communications surveys & tutorials, 18(2):1153–1176, 2015

    2015

  3. [11]

    A view on out-of-distribution identification from a statistical testing theory perspective

    Alberto Caron, Chris Hicks, and Vasilios Mavroudis. A view on out-of-distribution identification from a statistical testing theory perspective. arXiv preprint arXiv:2405.03052, 2024

    2024 arXiv

  4. [12]

    NODE-GAM: Neural generalized additive model for interpretable deep learning

    Chun-Hao Chang, Rich Caruana, and Anna Goldenberg. NODE-GAM: Neural generalized additive model for interpretable deep learning. In International Conference on Learning Representations, 2022

    2022

  5. [13]

    Explainable artificial intelligence for cybersecurity: a literature survey

    Fabien Charmet, Harry Chandra Tanuwidjaja, Solayman Ayoubi, Pierre-François Gimenez, Yufei Han, Houda Jmila, Gregory Blanc, Takeshi Takahashi, and Zonghua Zhang. Explainable artificial intelligence for cybersecurity: a literature survey. Annals of Telecommunications, 77(11):78...

    2022

  6. [14]

    The potential for artificial intelligence in healthcare

    Thomas Davenport and Ravi Kalakota. The potential for artificial intelligence in healthcare. Future healthcare journal, 6(2):94–98, 2019

    2019

  7. [15]

    Aleatory or epistemic? does it matter?Structural safety, 31(2):105–112, 2009

    Armen Der Kiureghian and Ove Ditlevsen. Aleatory or epistemic? does it matter?Structural safety, 31(2):105–112, 2009

    2009

  8. [16]

    Additive gaussian processes

    David K Duvenaud, Hannes Nickisch, and Carl Rasmussen. Additive gaussian processes. Advances in neural information processing systems, 24, 2011

    2011

  9. [17]

    Cyborg++: An enhanced gym for the development of autonomous cyber agents

    Harry Emerson, Liz Bates, Chris Hicks, and Vasilios Mavroudis. Cyborg++: An enhanced gym for the development of autonomous cyber agents. arXiv preprint arXiv:2410.16324, 2024

    2024 arXiv

  10. [18]

    Autonomous network defence using reinforce- ment learning

    Myles Foley, Chris Hicks, Kate Highnam, and Vasilios Mavroudis. Autonomous network defence using reinforce- ment learning. In Proceedings of the 2022 ACM on Asia Conference on Computer and Communications Security, pages 1252–1254, 2022

    2022

  11. [19]

    Inroads into autonomous network defence using explained reinforcement learning

    Myles Foley, Mia Wang, Chris Hicks, Vasilios Mavroudis, et al. Inroads into autonomous network defence using explained reinforcement learning. arXiv preprint arXiv:2306.09318, 2023

    2023 arXiv

  12. [20]

    Measuring network security using dynamic bayesian network

    Marcel Frigault, Lingyu Wang, Anoop Singhal, and Sushil Jajodia. Measuring network security using dynamic bayesian network. In Proceedings of the 4th ACM workshop on Quality of protection, pages 23–30, 2008

    2008

  13. [21]

    right to explanation

    Bryce Goodman and Seth Flaxman. European union regulations on algorithmic decision-making and a “right to explanation”. AI magazine, 38(3):50–57, 2017

    2017

  14. [22]

    On calibration of modern neural networks

    Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q Weinberger. On calibration of modern neural networks. In International conference on machine learning, pages 1321–1330. PMLR, 2017

    2017

  15. [23]

    Theory of disagreement-based active learning

    Steve Hanneke et al. Theory of disagreement-based active learning. Foundations and Trends® in Machine Learning, 7(2-3):131–309, 2014

    2014

  16. [24]

    Generalized additive models

    Trevor J Hastie. Generalized additive models. In Statistical models in S, pages 249–307. Routledge, 2017

    2017

  17. [25]

    Autonomous cyber defence: Beyond games? Turing Technical Report, 2024

    Chris Hicks and Vasilios Mavroudis. Autonomous cyber defence: Beyond games? Turing Technical Report, 2024. 12

    2024

  18. [26]

    Canaries and whistles: Resilient drone communication networks with (or without) deep reinforcement learning

    Chris Hicks, Vasilios Mavroudis, Myles Foley, Thomas Davies, Kate Highnam, and Tim Watson. Canaries and whistles: Resilient drone communication networks with (or without) deep reinforcement learning. In Proceedings of the 16th ACM Workshop on Artificial Intelligence and Securi...

    2023

  19. [27]

    What uncertainties do we need in bayesian deep learning for computer vision? Advances in neural information processing systems, 30, 2017

    Alex Kendall and Yarin Gal. What uncertainties do we need in bayesian deep learning for computer vision? Advances in neural information processing systems, 30, 2017

    2017

  20. [28]

    Simple and scalable predictive uncertainty estimation using deep ensembles

    Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. Simple and scalable predictive uncertainty estimation using deep ensembles. Advances in neural information processing systems, 30, 2017

    2017

  21. [29]

    Deep neural networks as gaussian processes

    Jaehoon Lee, Yasaman Bahri, Roman Novak, Samuel S Schoenholz, Jeffrey Pennington, and Jascha Sohl-Dickstein. Deep neural networks as gaussian processes. arXiv preprint arXiv:1711.00165, 2017

    2017 arXiv

  22. [30]

    The mythos of model interpretability: In machine learning, the concept of interpretability is both important and slippery

    Zachary C Lipton. The mythos of model interpretability: In machine learning, the concept of interpretability is both important and slippery. Queue, 16(3):31–57, 2018

    2018

  23. [31]

    A unified approach to interpreting model predictions

    Scott M Lundberg and Su-In Lee. A unified approach to interpreting model predictions. Advances in neural information processing systems, 30, 2017

    2017

  24. [32]

    Explaining network intrusion detection system using explainable ai framework

    Shraddha Mane and Dattaraj Rao. Explaining network intrusion detection system using explainable ai framework. arXiv preprint arXiv:2103.07110, 2021

    2021 arXiv

  25. [33]

    An adversarial approach for explainable ai in intrusion detection systems

    Daniel L Marino, Chathurika S Wickramasinghe, and Milos Manic. An adversarial approach for explainable ai in intrusion detection systems. In IECON 2018-44th Annual Conference of the IEEE Industrial Electronics Society, pages 3237–3243. IEEE, 2018

    2018

  26. [34]

    Adaptive webpage fingerprinting from tls traces

    Vasilios Mavroudis and Jamie Hayes. Adaptive webpage fingerprinting from tls traces. In 2023 53rd Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), pages 445–458. IEEE, 2023

    2023

  27. [35]

    Interpretable machine learning

    Christoph Molnar. Interpretable machine learning. Lulu. com, 2020

    2020

  28. [36]

    Priors for infinite networks

    Radford M Neal and Radford M Neal. Priors for infinite networks. Bayesian learning for neural networks, pages 29–53, 1996

    1996

  29. [37]

    Dynamic security risk management using bayesian attack graphs

    Nayot Poolsappasit, Rinku Dewri, and Indrajit Ray. Dynamic security risk management using bayesian attack graphs. IEEE Transactions on Dependable and Secure Computing, 9(1):61–74, 2011

    2011

  30. [38]

    A unifying view of sparse approximate gaussian process regression

    Joaquin Quinonero-Candela and Carl Edward Rasmussen. A unifying view of sparse approximate gaussian process regression. The Journal of Machine Learning Research, 6:1939–1959, 2005

    1939

  31. [39]

    Neural basis models for interpretability

    Filip Radenovic, Abhimanyu Dubey, and Dhruv Mahajan. Neural basis models for interpretability. Advances in Neural Information Processing Systems, 35:8414–8426, 2022

    2022

  32. [40]

    Uncertainty quantification and deep ensembles

    Rahul Rahaman et al. Uncertainty quantification and deep ensembles. Advances in Neural Information Processing Systems, 34:20063–20075, 2021

    2021

  33. [41]

    A survey of deep active learning

    Pengzhen Ren, Yun Xiao, Xiaojun Chang, Po-Yao Huang, Zhihui Li, Brij B Gupta, Xiaojiang Chen, and Xin Wang. A survey of deep active learning. ACM computing surveys (CSUR), 54(9):1–40, 2021

    2021

  34. [42]

    why should i trust you?

    Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. " why should i trust you?" explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pages 1135–1144, 2016

    2016

  35. [43]

    Toward optimal active learning through monte carlo estimation of error reduction

    Nicholas Roy and Andrew McCallum. Toward optimal active learning through monte carlo estimation of error reduction. Icml, williamstown, 2(441-448):4, 2001

    2001

  36. [44]

    Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

    Cynthia Rudin. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature machine intelligence, 1(5):206–215, 2019

    2019

  37. [45]

    Interpretable machine learning: Fundamental principles and 10 grand challenges

    Cynthia Rudin, Chaofan Chen, Zhi Chen, Haiyang Huang, Lesia Semenova, and Chudi Zhong. Interpretable machine learning: Fundamental principles and 10 grand challenges. Statistic Surveys, 16:1–85, 2022

    2022

  38. [46]

    Active learning literature survey.University of Wisconsin-Madison Department of Computer Sciences, 2009

    Burr Settles. Active learning literature survey.University of Wisconsin-Madison Department of Computer Sciences, 2009

    2009

  39. [47]

    Query by committee

    H Sebastian Seung, Manfred Opper, and Haim Sompolinsky. Query by committee. In Proceedings of the fifth annual workshop on Computational learning theory, pages 287–294, 1992

    1992

  40. [48]

    A value for n-person games

    Lloyd S Shapley. A value for n-person games. In Harold W. Kuhn and Albert W. Tucker, editors,Contributions to the Theory of Games II, pages 307–317. Princeton University Press, Princeton, 1953

    1953

  41. [49]

    Outside the closed world: On using machine learning for network intrusion detection

    Robin Sommer and Vern Paxson. Outside the closed world: On using machine learning for network intrusion detection. In 2010 IEEE symposium on security and privacy, pages 305–316. IEEE, 2010. 13

    2010

  42. [50]

    Entity-based reinforcement learning for autonomous cyber defence

    Isaac Symes Thompson, Alberto Caron, Chris Hicks, and Vasilios Mavroudis. Entity-based reinforcement learning for autonomous cyber defence. arXiv preprint arXiv:2410.17647, 2024

    2024 arXiv

  43. [51]

    An explainable machine learning framework for intrusion detection systems

    Maonan Wang, Kangfeng Zheng, Yanqing Yang, and Xiujuan Wang. An explainable machine learning framework for intrusion detection systems. IEEE Access, 8:73127–73141, 2020

    2020

  44. [52]

    Gaussian processes for machine learning, volume 2

    Christopher KI Williams and Carl Edward Rasmussen. Gaussian processes for machine learning, volume 2. MIT press Cambridge, MA, 2006

    2006

  45. [53]

    Using bayesian networks for cyber security analysis

    Peng Xie, Jason H Li, Xinming Ou, Peng Liu, and Renato Levy. Using bayesian networks for cyber security analysis. In 2010 IEEE/IFIP International Conference on Dependable Systems & Networks (DSN), pages 211–220. IEEE, 2010

    2010

  46. [54]

    Generalized out-of-distribution detection: A survey

    Jingkang Yang, Kaiyang Zhou, Yixuan Li, and Ziwei Liu. Generalized out-of-distribution detection: A survey. International Journal of Computer Vision, pages 1–28, 2024

    2024

  47. [55]

    Interpretable probabilistic bayesian neural networks for cybersecu- rity intrusion detection

    Tengfei Yang, Yuansong Qiao, and Brian Lee. Interpretable probabilistic bayesian neural networks for cybersecu- rity intrusion detection. IEEE Access, 2024

    2024

  48. [56]

    Towards trustworthy cybersecurity operations using bayesian deep learning to improve uncertainty quantification of anomaly detection

    Tengfei Yang, Yuansong Qiao, and Brian Lee. Towards trustworthy cybersecurity operations using bayesian deep learning to improve uncertainty quantification of anomaly detection. Computers & Security, page 103909, 2024

    2024

  49. [57]

    Gaussian process neural additive models

    Wei Zhang, Brian Barr, and John Paisley. Gaussian process neural additive models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 16865–16872, 2024. 14

    2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.