REVIEW 4 major objections 5 minor 57 references
Bridging Econometrics and AI: VaR Estimation via Reinforcement Learning and GARCH Models
T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash
Pith's one-line read Scaling GARCH-based Value-at-Risk by a reinforcement-learning risk classifier reduces breaches and capital charges while passing regulatory backtests on sixteen years of Euro Stoxx 50 data.
desk verdict A plausible hybrid GARCH-DDQN VaR model whose empirical gains depend on leak-free preprocessing; the paper needs revision to prove that and to fix overclaimed backtests. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
Three objects carry the argument. First, the risk threshold $c = \max\{r_{k+1} : r_{k+1} < \mathrm{VaR}_{k+1}(\alpha),\, k=1,\dots,H\}$, the mildest recent VaR violation over a horizon $H$, which converts returns into the binary label $y_t(c) = \mathbf{1}\{r_t \le c\}$ and ties the classification target directly to realized downside risk. Second, the Double Deep Q-Network classifier trained with an unbalanced reward scheme ($+1$ for true positives, $-1$ for false positives, $+\rho$ for true negatives, $-\rho$ for false negatives, with $\rho$ the minority-to-majority class ratio) so the agent learns to detect rare high-risk days rather than optimizing majority-class accuracy. Third, the adjustment rule $\mathrm{VaR}^{ML}(\alpha) = \kappa\cdot\mathrm{VaR}_{t+1}(\alpha)$ with $\kappa = 1-b_1$ for low-risk predictions and $\kappa = 1+b_2$ for high-risk predictions, with $b_1 = 0.30$, $b_2 = 0.20$ fixed by grid search and corroborated by Bayesian posterior estimation. Because the adjustment is a positive scalar multiple of a coherent risk measure, the paper argues that $\mathrm{VaR}^{ML}$ inherits VaR's homogeneity, monotonicity, and translation properties.
What would settle it
Re-run the pipeline twice, once with Boruta, Min-Max, and ADASYN fitted only on the training window and once with them fitted on the full sample, and compare the test-split accuracy and the Kupiec/Christoffersen statistics; if the second run beats the first, the reported results are look-ahead artifacts. As a control, replace the DDQN labels in the adjustment rule with random labels and check whether the violation counts and backtest decisions stay unchanged, which would show the classifier itself, rather than the $b_1,b_2$ scaling, is doing the work.
Extended reading notes
Core claim
The central claim is that the Classification-Adjusted Value-at-Risk, defined as $\mathrm{VaR}^{ML}(\alpha) = (1-b_1)\cdot\mathrm{VaR}_{t+1}(\alpha)$ in predicted low-risk states and $(1+b_2)\cdot\mathrm{VaR}_{t+1}(\alpha)$ in predicted high-risk states, strictly improves on the underlying GARCH or GJR-GARCH measure: it achieves the correct violation frequency, restores the independence of violations that the plain econometric models violate, and does so with less conservative capital buffers. The supporting record is the backtest comparison: on the validation window the RL-adjusted GARCH VaR produces 52 violations at the 5% level against 73 for the original model and passes both the Kupiec and Christoffersen tests that reject the original, while on the test window violations stay close to their expected counts (52 actual versus 50 expected at 5%) and a Wilcoxon rank-sum test against GARCH yields $p = 0.0039$. The authors also report that exceedances over the adjusted threshold remain consistent with a Generalized Pareto Distribution, which they read as evidence that tail behavior is preserved, and that the adjusted VaR lies significantly above the GARCH VaR under a Mann-Whitney test, which they interpret as lower capital charges at unchanged regulatory coverage.
Load-bearing premise
The load-bearing premise is that all preprocessing, namely Boruta feature selection, Min-Max normalization, and ADASYN oversampling, is fitted only inside the training split so that the validation and test samples are never seen by any preprocessing step; the paper never states this, yet the reported 79.4% test accuracy and the out-of-sample backtest pass rates depend on it.
Editorial extensions
If this is right
- If the framework is correct, a risk manager can report a VaR that is less conservative in calm periods and more conservative in stressed ones without failing the standard backtests, since the adjusted model is accepted by both the Kupiec and Christoffersen tests at the 5% and 1% levels.
- The reported Mann-Whitney results imply the adjusted VaR sits systematically above the GARCH VaR distribution, so capital charges fall at unchanged coverage; this is the paper's economic payoff.
- Because the paper argues that better VaR estimates directly improve the Expected Shortfall tail estimate, the same classification step carries over to the Basel III framework that replaced VaR with ES.
- Because inference takes under 0.1 seconds per step while training is a one-off cost, the adjusted VaR can be recomputed daily in an operational risk-monitoring system.
Reading between the lines
- The authors validate on a single index, Euro Stoxx 50; a natural test of transferability is to re-run the identical DDQN-plus-GARCH pipeline on US equities, FX, or commodities and check whether the 79.4% test accuracy and the violation reductions survive.
- Their finding of a negative posterior correlation between $b_1$ and $b_2$ suggests the benefit comes from the balance of the two adjustments, not from either alone; a one-parameter variant would presumably sacrifice either coverage or capital efficiency, which could be tested directly.
- The threshold $c$ is defined from realized VaR violations, which creates a feedback loop between the classifier's target and the backtest statistic; one could profit from this loop deliberately by training the agent with a reward proportional to breach size or to capital at risk, optimizing the economic objective rather than label accuracy.
- One can separate the contributions of the classifier and the adjustment by setting $b_1 = b_2 = 0$ (no adjustment) and checking whether violation counts already improve; the paper does not report this ablation, but it would establish how much of the gain comes from the scaling rule versus the regime labels themselves.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a hybrid Value-at-Risk (VaR) framework that combines GARCH/GJR-GARCH volatility forecasts with a Double Deep Q-Network (DDQN) reinforcement learning classifier. The classifier assigns each day to a low-risk or high-risk state based on a threshold derived from past VaR violations, and the GARCH-based VaR is then scaled by factors (1-b1) or (1+b2) depending on the predicted state. Empirical validation on daily Euro Stoxx 50 data from September 2008 to March 2025 reports 79.4% test classification accuracy, reduced numbers of VaR violations relative to GARCH/GJR-GARCH and TCN baselines, passing Kupiec and Christoffersen backtests, consistency with Extreme Value Theory tail fits, and lower capital requirements while maintaining regulatory coverage. The paper argues that the adaptive, RL-based adjustment yields both statistically valid and capital-efficient risk measures.
Significance. If the empirical claims are correct, the paper offers a practically relevant way to combine econometric volatility modeling with deep reinforcement learning for regulatory risk measurement, with clear economic benefits through lower capital charges without sacrificing coverage. The paper has notable strengths: a long (over 16-year) test dataset spanning crisis episodes, comparison against several baselines (LR, SVM, ANN, MLP, TCN), use of class-imbalance metrics beyond accuracy (G-Mean, recall), a detailed appendix with hyperparameters, and explicit backtesting and EVT robustness checks. However, the current manuscript does not establish that the entire empirical pipeline is leak-free, and several central claims are either circular or contradicted by the paper's own tables. The significance is therefore conditional: the contribution would be substantial after the methodological gaps are resolved, but as written the key evidence is not reliable.
major comments (4)
- [Appendix D.1 and Section 3.1] The manuscript does not establish that the Boruta feature selection, Min-Max normalization, and ADASYN oversampling are confined to the training split. Appendix D.1 states that 'The features selected using the Boruta algorithm were split chronologically into training, validation and test sets to avoid anticipation bias,' which reads as feature selection on the full sample followed by splitting; the same appendix then states that 'all predictors were rescaled to [0,1] using Min–Max normalization' and that ADASYN addressed class imbalance, without specifying that the normalization parameters and the ADASYN fit are training-only. Section 3.1 similarly describes the chronological split but then discusses Boruta selection without an explicit statement that the selection is performed only on the training portion. Because the 79.4% test accuracy and the test-sample violation reductions (e.g., from 64 to 52 in Table 9) are the principal evidence for outperformance, this ambiguity is load-bearing. The authors must state precisely which preprocessing steps are fit on which split, and if the pipeline as run used any full-sample information, the experiments must be rerun with strictly training-only preprocessing.
- [Section 4.1 and Tables 7/8] The validation-sample backtest results are not independent evidence of model quality because the adjustment parameters b1 and b2 are deliberately calibrated to minimize the number of VaR violations on the validation set (Section 4.1: 'the pair (b1 = 0.30, b2 = 0.20) produces the lowest number of exceedances'). Presenting the validation reduction in breaches (Tables 6-9) as evidence of outperformance is circular. At the same time, the text in Section 3.3 claims that 'RL-adjusted VaR models pass through validation and test samples' and 'show no systematic bias,' which is directly contradicted by Table 8(a): the RL GARCH 5% validation Kupiec test reports p = 0.032 with decision 'RejectH0'. Please correct the text, and either remove the validation-based claims or explicitly re-frame them as in-sample calibration diagnostics rather than out-of-sample evidence.
- [Section 2.3, Eq. (9)] The classification threshold c is defined as the maximum of returns that breach VaR over a horizon H, but the manuscript never states whether H is restricted to the training period or whether c is recomputed dynamically in a way that avoids lookahead. If the threshold is computed using validation- or test-period returns (or using VaR estimates that incorporate those returns), then the target labels y_t(c) in the test sample are constructed with information from the test period itself, which would inflate the reported 79.4% accuracy and the out-of-sample backtest pass rates. The authors must specify the exact construction window for c, demonstrate that the labels in the validation and test splits are generated without using those splits' returns, and discuss the endogeneity that arises because the labels are defined through violations of the very GARCH VaR model that is later adjusted.
- [Section 2.4, Proposition 2.1] The proof of Proposition 2.1 does not establish that Classification-Adjusted VaR preserves the fundamental properties of Value-at-Risk. In particular, translation invariance fails: for a portfolio with a deterministic cash position c, VaR_ML(X + c) = κ·(VaR(X) + c) = κ·VaR(X) + κ·c, which is not equal to VaR_ML(X) + c unless κ = 1. The claim that 'subadditivity and translation invariance are unaffected' is asserted without proof and is not a consequence of positive homogeneity. Either the proposition and proof must be corrected, or the authors should state the weaker and correct result that VaR_ML is a positive-homogeneous, monotonic transformation of the baseline VaR.
minor comments (5)
- [Section 5] The phrase 'captures non-nonlinearities' in the conclusion is a typo and should read 'nonlinearities.'
- [Section 3.3] The claim that the RL model's skewness and kurtosis are 'far closer to empirical distributions' would be more convincing if the empirical skewness and kurtosis of the test-period returns were reported next to the model values; please add these statistics.
- [Table 2] There is a formatting issue in the features table: 'sig gjr-GARCH0.013' is missing a space between the variable name and its value; please fix the alignment.
- [Appendix D.1.1] The sentence beginning 'For time series models, ARIMA orders capture the optimal trade-off...' has a subject-verb agreement issue and is difficult to parse; please rewrite for clarity.
- [Section 3.1] The paper reports percentages of low- and high-risk observations in each split but does not give the absolute numbers of observations in the training, validation, and test sets; please add these to Section 3.1 or Table 1.
Circularity Check
The validation violation reduction is a fitted outcome: (b1,b2) are chosen to minimize validation violations and the same validation counts/backtests are then reported as evidence of outperformance. The threshold and adjusted VaR are defined from the same GARCH VaR, entangling the violation-reduction claim with the construction.
-
fitted input called prediction
[Section 4.1 (Calibration) and Section 4.3 / Table 9]
"The cross-validation approach explores a grid of (b1,b2) combinations and evaluates their performance in terms of the number of VaR violations across validation windows. As illustrated in Figure 9a, the violation surface shows a clear minimum, indicating that the pair (b1 = 0.30,b2 = 0.20) produces the lowest number of exceedances. [...] These findings are further supported by the aggregated violation counts reported in Table 9, where the RL-adjusted VaR consistently reports fewer exceedances than both GARCH-type and TCN models, particularly at the 5% risk level."
The pair (b1,b2) is selected on the validation split by minimizing the number of VaR violations, as stated explicitly in Section 4.1. The same validation split is then used in Table 9 and in the backtest Tables 7–8 as evidence that the RL-adjusted VaR 'consistently reports fewer exceedances' and passes Kupiec/Christoffersen tests. For the validation sample, the reduction in violations is the objective function of the grid search, so the reported improvement over GARCH/GJR is a restatement of the calibration criterion, not an out-of-sample prediction. Only the test-sample violation counts are independent of this particular fit.
-
self definitional
[Section 2.3, Eq. (9)-(10); Section 2.4, Eq. (11)]
"c = max { r_{k+1} | r_{k+1} < VaR_{k+1}(α), k = 1,...,H } ... This formulation explicitly links the classification threshold to the realized downside risk, aligning the directional prediction with the practical objective of reducing VaR violations. ... VaRML(α) = (1−b1)·VaRt+1(α), if r̂t+1 = 'Low risk'; (1 +b2)·VaRt+1(α), if r̂t+1 = 'High risk'."
The binary target y_t(c)=1{r_t≤c} is constructed from c, the mildest observed GARCH VaR violation (Eq. 9), and the adjusted VaR (Eq. 11) is a rescaling of that same GARCH VaR by (1−b1) or (1+b2) according to the predicted class. Thus the class labels, the adjustment multipliers, and the reported VaR are all functions of the same baseline GARCH VaR process. The validation 'reduction in breaches' is therefore not a comparison of the hybrid model against an independent risk model; it is a relabeling of the baseline's own violation days, tuned by b1,b2. The test classifier accuracy (79.4%) is independent, but the violation-reduction claim is entangled with the definition.
full rationale
The paper's clearest circularity is the calibration of the VaR adjustment parameters on the validation violation count and the subsequent use of validation violation/backtest statistics as evidence of outperformance. Section 4.1 states that (b1,b2) are chosen to minimize the number of validation violations, and Section 4.3/Table 9 then presents those same validation counts as supporting the claim that RL-adjusted VaR reduces exceedances. That is a fitted input presented as a prediction. Additionally, the threshold c is defined from GARCH VaR violations and the adjusted VaR is a multiplicative transform of the same GARCH VaR, so the validation backtest is assessing a quantity that was optimized against the baseline's own violation process. The test-sample results and the 79.4% classification accuracy are genuinely out-of-sample and give the central claim independent content; there is no load-bearing self-citation or imported uniqueness theorem. The score is 6 rather than higher because the test split and the classifier evaluation are not fitted to the violation count, so the central contribution retains independent empirical support.
Assumptions & free parameters
free parameters (5)
- b1 =
0.30
- b2 =
0.20
- Threshold horizon H =
not specified
- Student-t degrees of freedom nu =
not reported
- DDQN hyperparameters =
gamma=0.95, epsilon-decay=0.995, lr=0.0005, layers (96,64)
assumptions (6)
- domain assumption Returns follow a conditional location-scale model r_t = mu_t + sigma_t z_t with i.i.d. standardized innovations (Eq. 3).
- domain assumption GARCH and GJR-GARCH variance processes satisfy stationarity and positivity constraints (Section 2.2).
- domain assumption The threshold c defined by Eq. (9) produces a binary risk label that is learnable from the chosen market features.
- domain assumption The reward design in Eq. (12) with scaling factor rho aligns the RL policy with the goal of minimizing VaR violations.
- domain assumption The chronological data split is leak-free: Boruta feature selection, Min-Max normalization, and ADASYN oversampling are applied only on the training partition.
- standard math Pickands-Balkema-de Haan theorem: exceedances over a high threshold follow a Generalized Pareto Distribution.
invented entities (1)
-
Classification-Adjusted VaR (VaR_ML)
Cite this review
Pith. "Pith review of Bridging Econometrics and AI: VaR Estimation via Reinforcement Learning and GARCH Models." pith.science (2026). https://pith.science/paper/IIM3C45B
@misc{pith2026250416635,
author = {Pith},
title = {Pith review of: Bridging Econometrics and AI: VaR Estimation via Reinforcement Learning and GARCH Models},
year = {2026},
howpublished = {\url{https://pith.science/paper/IIM3C45B}},
note = {Machine review of arXiv:2504.16635}
}
read the original abstract
In an environment of increasingly volatile financial markets, the accurate estimation of risk remains a major challenge. Traditional econometric models, such as GARCH and its variants, are based on assumptions that are often too rigid to adapt to the complexity of the current market dynamics. To overcome these limitations, we propose a hybrid framework for Value-at-Risk (VaR) estimation, combining GARCH volatility models with deep reinforcement learning. Our approach incorporates directional market forecasting using the Double Deep Q-Network (DDQN) model, treating the task as an imbalanced classification problem. This architecture enables the dynamic adjustment of risk-level forecasts according to market conditions. Empirical validation on daily Eurostoxx 50 data covering periods of crisis and high volatility shows a significant improvement in the accuracy of VaR estimates, as well as a reduction in the number of breaches and also in capital requirements, while respecting regulatory risk thresholds. The ability of the model to adjust risk levels in real time reinforces its relevance to modern and proactive risk management.
Figures
Figures from the paper (11 more)
Reference graph
Works this paper leans on
-
[1]
H. Alostad and H. Davulcu. Directional prediction of stock prices using breaking news on twitter. In Web Intelligence, volume 15, pages 1--17. IOS Press, 2017
work page 2017
-
[2]
E. K. Ampomah, Z. Qin, and G. Nyame. Evaluation of tree-based ensemble machine learning models in predicting stock price direction of movement. Information, 11 0 (6): 0 332, 2020
work page 2020
-
[3]
R. Barandela, R. M. Valdovinos, J. S. S \'a nchez, and F. J. Ferri. The imbalanced training sample problem: Under or over sampling? In Structural, Syntactic, and Statistical Pattern Recognition: Joint IAPR International Workshops, SSPR 2004 and SPR 2004, Lisbon, Portugal, August 18-20, 2004. Proceedings, pages 806--814. Springer, 2004
work page 2004
-
[4]
T. Bollerslev. Generalized autoregressive conditional heteroskedasticity. Journal of econometrics, 31 0 (3): 0 307--327, 1986
work page 1986
-
[5]
N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer. Smote: synthetic minority over-sampling technique. Journal of artificial intelligence research, 16: 0 321--357, 2002
2002
-
[6]
J. Chung and Y. Hong. Model-free evaluation of directional predictability in foreign exchange markets. Journal of Applied Econometrics, 22 0 (5): 0 855--889, 2007
work page 2007
-
[7]
T. Cui, S. Ding, H. Jin, and Y. Zhang. Portfolio constructions in cryptocurrency market: A cvar-based deep reinforcement learning approach. Economic Modelling, 119: 0 106078, 2023
work page 2023
-
[8]
S. Dreiseitl and L. Ohno-Machado. Logistic regression and artificial neural network classification models: a methodology review. Journal of biomedical informatics, 35 0 (5-6): 0 352--359, 2002
work page 2002
Show all 57 references
-
[9]
Elhassan and M
T. Elhassan and M. Aljurf. Classification of imbalance data using tomek link (t-link) combined with random under-sampling (rus) as a data reduction method. Global J Technol Optim S, 1: 0 2016, 2016
2016
-
[10]
Fawcett and F
T. Fawcett and F. Provost. Adaptive fraud detection. Data mining and knowledge discovery, 1 0 (3): 0 291--316, 1997
1997
-
[11]
Firdous, N
N. Firdous, N. M. U. Din, and A. Assad. An imbalanced classification approach for establishment of cause-effect relationship between heart-failure and pulmonary embolism using deep reinforcement learning. Engineering Applications of Artificial Intelligence, 126: 0 107004, 2023
2023
-
[12]
W. R. Gilks, S. Richardson, and D. Spiegelhalter. Markov chain Monte Carlo in practice. CRC press, 1995
1995
-
[13]
L. R. Glosten, R. Jagannathan, and D. E. Runkle. On the relation between the expected value and the volatility of the nominal excess return on stocks. The journal of finance, 48 0 (5): 0 1779--1801, 1993
1993
-
[14]
Goodfellow, Y
I. Goodfellow, Y. Bengio, A. Courville, and Y. Bengio. Deep learning, volume 1. MIT press Cambridge, 2016
2016
-
[15]
Guidotti, A
R. Guidotti, A. Monreale, S. Ruggieri, F. Turini, F. Giannotti, and D. Pedreschi. A survey of methods for explaining black box models. ACM computing surveys (CSUR), 51 0 (5): 0 1--42, 2018
2018
-
[16]
Han, W.-Y
H. Han, W.-Y. Wang, and B.-H. Mao. Borderline-smote: a new over-sampling method in imbalanced data sets learning. In International conference on intelligent computing, pages 878--887. Springer, 2005
2005
-
[17]
H. He, Y. Bai, E. A. Garcia, and S. Li. Adasyn: Adaptive synthetic sampling approach for imbalanced learning. In 2008 IEEE international joint conference on neural networks (IEEE world congress on computational intelligence), pages 1322--1328. Ieee, 2008
2008
-
[18]
A. Kanas. Neural network linear forecasts for stock returns. International Journal of Finance & Economics, 6 0 (3): 0 245--254, 2001
2001
-
[19]
Kotsiantis, D
S. Kotsiantis, D. Kanellopoulos, P. Pintelas, et al. Handling imbalanced datasets: A review. GESTS international transactions on computer science and engineering, 30 0 (1): 0 25--36, 2006
2006
-
[20]
M. B. Kursa and W. R. Rudnicki. Feature selection with the boruta package. Journal of statistical software, 36: 0 1--13, 2010
2010
-
[21]
M. B. Kursa, A. Jankowski, and W. R. Rudnicki. Boruta--a system for feature selection. Fundamenta Informaticae, 101 0 (4): 0 271--285, 2010
2010
-
[22]
C. Lea, R. Vidal, A. Reiter, and G. D. Hager. Temporal convolutional networks: A unified approach to action segmentation. In Computer Vision--ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 15-16, 2016, Proceedings, Part III 14, pages 47--54. Springer, 2016
2016
-
[23]
Linton and Y.-J
O. Linton and Y.-J. Whang. The quantilogram: With an application to evaluating directional predictability. Journal of Econometrics, 141 0 (1): 0 250--282, 2007
2007
-
[24]
S. M. Lundberg and S.-I. Lee. A unified approach to interpreting model predictions. Advances in neural information processing systems, 30, 2017
2017
-
[25]
V. Mnih, K. Kavukcuoglu, D. Silver, A. Graves, I. Antonoglou, D. Wierstra, and M. Riedmiller. Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602, 2013
2013 arXiv
-
[26]
Morimura, M
T. Morimura, M. Sugiyama, H. Kashima, H. Hachiya, and T. Tanaka. Parametric return density estimation for reinforcement learning. arXiv preprint arXiv:1203.3497, 2012
2012 arXiv
-
[27]
F. Murtagh. Multilayer perceptrons for classification and regression. Neurocomputing, 2 0 (5-6): 0 183--197, 1991
1991
-
[28]
Nevasalmi
L. Nevasalmi. Forecasting multinomial stock returns using machine learning methods. The Journal of Finance and Data Science, 6: 0 86--106, 2020
2020
-
[29]
Noorani, C
E. Noorani, C. N. Mavridis, and J. S. Baras. Risk-sensitive reinforcement learning with exponential criteria. IEEE Transactions on Cybernetics, 2025
2025
-
[30]
H. Nyberg. Forecasting the direction of the us stock market with dynamic binary probit models. International Journal of Forecasting, 27 0 (2): 0 561--578, 2011
2011
-
[31]
B. C. on Banking Supervision. Basel iii: Finalising post-crisis reforms. Bank for International Settlements, 2017
2017
-
[32]
R. M. Pereira, Y. M. Costa, and C. N. Silla Jr. Mltl: A multi-label approach for the tomek link undersampling algorithm. Neurocomputing, 383: 0 95--105, 2020
2020
-
[33]
Pokou, J
F. Pokou, J. Sadefo Kamdem, and F. Benhmad. Hybridization of arima with learning models for forecasting of stock market time series. Computational Economics, 63 0 (4): 0 1349--1399, 2024
2024
-
[34]
F. V. M. Pokou. Une contribution sur l’allocation ou la pr \'e vision d’actifs d’un portefeuille . PhD thesis, Universit \'e de Montpellier, 2022
2022
-
[35]
M. L. Puterman. Markov decision processes: discrete stochastic dynamic programming. John Wiley & Sons, 2014
2014
-
[36]
why should i trust you?
M. T. Ribeiro, S. Singh, and C. 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
-
[37]
Rosenblatt
F. Rosenblatt. The perceptron: a probabilistic model for information storage and organization in the brain. Psychological review, 65 0 (6): 0 386, 1958
1958
-
[38]
C. Rudin. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature machine intelligence, 1 0 (5): 0 206--215, 2019
2019
-
[39]
Stanko and K
S. Stanko and K. Macek. Risk-averse distributional reinforcement learning: A cvar optimization approach. In IJCCI, pages 412--423, 2019
2019
-
[40]
M. A. I. Sunny, M. M. S. Maswood, and A. G. Alharbi. Deep learning-based stock price prediction using lstm and bi-directional lstm model. In 2020 2nd novel intelligent and leading emerging sciences conference (NILES), pages 87--92. IEEE, 2020
2020
-
[41]
Supervision
B. Supervision. Basel committee on banking supervision. Principles for Sound Liquidity Risk Management and Supervision (September 2008), 2011
2008
-
[42]
Supervision
B. Supervision. Basel committee on banking supervision. 2012
2012
-
[43]
Sutton and A
R. Sutton and A. Barto. Reinforcement learning: An introduction. IEEE Transactions on Neural Networks, 9 0 (5): 0 1054--1054, 1998
1998
-
[44]
Tang, Y.-Q
Y. Tang, Y.-Q. Zhang, N. V. Chawla, and S. Krasser. Svms modeling for highly imbalanced classification. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 39 0 (1): 0 281--288, 2008
2008
-
[45]
A. Tealab. Time series forecasting using artificial neural networks methodologies: A systematic review. Future Computing and Informatics Journal, 3 0 (2): 0 334--340, 2018
2018
-
[46]
I. Tomek. Two modifications of cnn. 1976
1976
-
[47]
Van Hasselt, A
H. Van Hasselt, A. Guez, and D. Silver. Deep reinforcement learning with double q-learning. In Proceedings of the AAAI conference on artificial intelligence, volume 30, 2016
2016
-
[48]
V. Vapnik. The nature of statistical learning theory. Springer science & business media, 1999
1999
-
[49]
C. J. C. H. Watkins. Learning from delayed rewards. 1989
1989
-
[50]
C. C. White III and D. J. White. Markov decision processes. European Journal of Operational Research, 39 0 (1): 0 1--16, 1989
1989
-
[51]
D. L. Wilson. Asymptotic properties of nearest neighbor rules using edited data. IEEE Transactions on Systems, Man, and Cybernetics, 0 (3): 0 408--421, 2007
2007
-
[52]
Yang, X.-Y
H. Yang, X.-Y. Liu, S. Zhong, and A. Walid. Deep reinforcement learning for automated stock trading: An ensemble strategy. In Proceedings of the first ACM international conference on AI in finance, pages 1--8, 2020
2020
-
[53]
C. Ying, X. Zhou, H. Su, D. Yan, N. Chen, and J. Zhu. Towards safe reinforcement learning via constraining conditional value-at-risk. arXiv preprint arXiv:2206.04436, 2022
2022 arXiv
-
[54]
M. Zeng, B. Zou, F. Wei, X. Liu, and L. Wang. Effective prediction of three common diseases by combining smote with tomek links technique for imbalanced medical data. In 2016 IEEE International Conference of Online Analysis and Computing Science (ICOACS), pages 225--228. IEEE, 2016
2016
-
[55]
G. P. Zhang. Time series forecasting using a hybrid arima and neural network model. Neurocomputing, 50: 0 159--175, 2003
2003
-
[56]
Zhang and M
H. Zhang and M. Li. Rwo-sampling: A random walk over-sampling approach to imbalanced data classification. Information Fusion, 20: 0 99--116, 2014
2014
-
[57]
Zhang, S
Q. Zhang, S. Leng, X. Ma, Q. Liu, X. Wang, B. Liang, Y. Liu, and J. Yang. Cvar-constrained policy optimization for safe reinforcement learning. IEEE transactions on neural networks and learning systems, 36 0 (1): 0 830--841, 2024
2024
Reviewed August 16, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.