betaFeedback

WorksComputer Science

Convex-SKM-Net: Low-Latency Block Halfspace Corrections for Neural Convex Constraint Satisfaction

Abstract

Fast inference and strict constraint satisfaction are essential for learning-based decision-making. Existing methods often rely on constraint-specific projections or feasible-set parameterizations, while iterative repair can be costly. We introduce Convex-SKM-Net, a convex constraint-satisfaction layer that forms linearization halfspaces from residual and (sub)gradient oracles without constraint-specific inequality projectors. It applies fixed-depth dual projected-gradient descent (dual PGD) to minimum-displacement block problems using GPU-friendly matrix operations. We jointly learn the predictor and a positive-definite mobility, adapting the correction geometry to improve objective quality. We prove convergence of the outer iterates at fixed dual depth and linear convergence in distance under Slater's condition and bounded subgradients. All our test outputs on three second-order-cone model-predictive control (SOC-MPC) horizons and six power-system convex relaxations satisfy a 10−310^{-3} feasibility tolerance. Comparisons with the latest constraint-satisfaction methods show inference speedups of up to 144.3×144.3\times on SOC-MPC and 7.7×7.7\times on a convex quadratically constrained quadratic programming (QCQP) benchmark with joint training. On QCQP, the mean objective gap is 0.012%0.012\%, with full test feasibility at 10−410^{-4}. Learning mobility reduces the five-seed affine-task mean gap by 22.4%22.4\% and improves gaps in matched power-system comparisons.

Introduction

Learning-based decision-making in power systems, model-predictive control, and wireless resource allocation requires fast predictions that satisfy physical constraints and retain good objective values [12, 7, 2]. Neural predictions can violate constraints even when violations are penalized during training [33, 12]. Constraint layers enforce feasibility through feasible-set parameterizations, optimization, or iterative repair. Their use depends on the operations available for each constraint family: parameterizations may require interior points [31], and splitting methods require component projections [16, 30]. Residual-based repair works directly with constraint values and derivatives, but repeated correction can remain costly [12, 27, 8]. Repair also changes the predicted decision, so its direction affects the final objective. We therefore design both the correction computation and its directional cost.

Convex-SKM-Net uses convexity to give different inequalities a common local representation. A residual and a subgradient define a linearization halfspace containing the feasible set, allowing constraints to be handled without family-specific projectors. After satisfying the affine equalities, we formulate minimum-displacement problems over selected blocks of these halfspaces. Fixed-depth dual projected-gradient descent (dual PGD) computes block updates with GPU-friendly matrix operations, without per-block factorizations or linear solves. Several block corrections reuse one linearization. On second-order-cone model-predictive control (SOC-MPC), inference is 80.7–144.3× faster than fixed-400-step Πnet. On the 300-bus semidefinite relaxation, postprocessing a fully finetuned predictor is 7.7× faster than LMI-Net, with a lower mean gap on identical raw predictions.

We also learn how displacement is measured during repair. Euclidean distance assigns the same cost to equal-length moves, although their effects on the task objective can differ. A positive-definite mobility changes these directional costs while preserving the feasible set. Joint training adjusts both the predictor’s initial output and the geometry of its subsequent corrections. The same layer supports postprocessing with a prescribed mobility, including the identity. Learning mobility reduces the five-seed mean gap on the affine task by 22.4%. In matched joint-training comparisons, it also reduces mean gaps by 11.0% and 8.1% on the 300-bus conic and semidefinite relaxations. Our contributions are:

  1. General convex constraint satisfaction. We introduce a layer using residual and (sub)gradient oracles without constraint-specific inequality projectors, supporting both postprocessing and joint training.

  1. Block corrections with dual PGD. Our GPU-friendly block updates use fixed dual depth without per-block factorizations. We establish convergence of the outer iterates and linear convergence in distance under Slater’s condition and bounded subgradients.

  1. Learnable mobility. We learn directional correction costs jointly with the predictor, improving objective gaps on affine and nonlinear convex tasks.

Related Work

We review neural constraint layers and the numerical methods underlying our corrections.

Feasible-set parameterization. Linear-constraint constructions use polyhedral generators, gauge maps, or decision rules ([14, 21, 9]); other approaches calibrate constraints or restrict network maps over prescribed input regions ([4] (Zhao et al. 2023)). RAYEN uses an interior point and ray boundaries, computing the point offline for fixed constraints ([31]). Radial transformations, invertible maps, and interior-point predictions also construct constrained outputs ([19, 23, 22, 28]).

Differentiable solver layers. Affine repair uses closed-form updates or subconstraint decompositions ([24] (Zhao et al. 2026)). Optimization layers differentiate through constrained solutions during training ([3, 1, 24]). Linear constraints admit generalized Sinkhorn and entropy-regularized dual iterations, with linear repair also used in tabular generative models ([32, 33, 29]). Πnet and LMI-Net use component projections in operator splitting ([16, 30]); Dykstra’s algorithm and alternating projections also support constrained prediction ([7, 10]).

Among residual-based layers, DC3 combines equality completion with violation gradients, and FS-Net minimizes violation ([12, 27]). Other updates use adaptive repair, damped linearization, optimality conditions, or regularized Newton steps ([20, 15, 17, 8]).

Sampling Kaczmarz–Motzkin (SKM) corrects the most violated sampled row ([11]); nonlinear SKM handles nonlinear equations and variants with convex constraints ([34]). T-SKM-Net unrolls row repair; AT-SKM-Net uses graph-guided sampling for dynamic linear constraints ([35] (Zhu et al. 2026)). Linear block methods project onto equalities or fit boundaries ([26, 36, 34, 6]); subgradient, parallel, block-iterative, and minibatch methods also address convex inequalities ([13, 18, 5, 25]).

Method: Convex-SKM-Net

Overview. Figure 1 illustrates the layer ycorr=R(x,y^;P)y_{\mathrm{corr}}=\mathcal{R}(x,\hat{y};P), which takes an instance xx, a prediction y^\hat{y}, and a prescribed positive-definite mobility PP. It first satisfies the affine equalities, then restricts corrections to their nullspace. Each outer iteration builds linearization halfspaces from residuals and (sub)gradients. The layer selects a block using normalized violation scores and applies TT dual PGD steps to its minimum-displacement problem. It performs SS block corrections before refreshing the linearization. Mobility controls normalization and directional costs and stays fixed during one repair. A coordinate transformation reuses the Euclidean dual solver. Algorithm 1 gives the full procedure. Postprocessing supplies PP directly; joint training learns its parameters together with the predictor.

Figure 1. Repair in reduced space with fixed (blue) or learned (red) mobility. The inset shows block reselection under one linearization. Trajectories are schematic.

For input xx, a predictor returns y^=fθ(x)∈Rn\hat{y}=f_{\theta}(x)\in\mathbb{R}^{n}. Assume the nonempty feasible set Y(x)={y:C(x)y=d(x), gi(y,x)≤0, i=1,…,m}\mathcal{Y}(x)=\{y:C(x)y=d(x),\ g_i(y,x)\leq0,\ i=1,\ldots,m\} has closed convex inequalities with queryable residuals and subgradients, and projectable affine equalities. Set yeq=P{Cy=d}(y^)y_{\mathrm{eq}=P_{\{Cy=d\}}(\hat{y})} and take an orthonormal basis NN of ker⁡C\ker C, so CN=0CN=0 and NTN=IqN^{\mathsf{T}}N=I_q. Then

y=yeq+Nz,hi(z)=gi(yeq+Nz,x),Ω={z:hi(z)≤0, ∀i}.(1)y=y_{\mathrm{eq}}+Nz,\qquad h_i(z)=g_i(y_{\mathrm{eq}}+Nz,x),\qquad\Omega=\{z:h_i(z)\leq0,\ \forall i\}. \tag*{(1)}

Input dependence is omitted below. Appendix § 12 gives the residual and (sub)gradient oracles for each task.

Halfspace block corrections

At zkz^k, take aik∈∂hi(zk)a_i^k\in\partial h_i(z^k). Convexity gives hi(z)≥ℓik(z)=hi(zk)+⟨aik,z−zk⟩h_i(z)\geq\ell_i^k(z)=h_i(z^k)+\langle a_i^k,z-z^k\rangle. Hence the linearization halfspace contains the feasible set, with single-row SKM correction

Hik={z:ℓik(z)≤0},Ω⊆Hik,Δik=−[hi(zk)]+∥aik∥2aik.(2)H_i^k=\{z:\ell_i^k(z)\leq0\},\qquad\Omega\subseteq H_i^k,\qquad\Delta_i^k=-\frac{[h_i(z^k)]_+}{\lVert a_i^k\rVert^2}a_i^k. \tag*{(2)}

Satisfied rows make no update; violated rows have nonzero subgradients (Lemma A.1). We apply this construction to a block of constraints.

For nonzero rows define

uik=aik∥aik∥,βik=(aik)Tzk−hi(zk)∥aik∥,u_i^k=\frac{a_i^k}{\lVert a_i^k\rVert},\qquad\beta_i^k=\frac{(a_i^k)^{\mathsf{T}}z^k-h_i(z^k)}{\lVert a_i^k\rVert},

so Hik={z:(uik)⊤z≤βik}H_i^k = \{z : (u_i^k)^\top z \le\beta_i^k\}. Zero-subgradient rows use zero placeholders. At inner iterate ww, initially zkz^k, select the bb largest normalized violation scores δik(w)=[(uik)⊤w−βik]+\delta_i^k(w) = [(u_i^k)^\top w - \beta_i^k]_+, with 1≤b≤m1 \le b \le m. Normalization removes positive row scaling; zero-score rows fill the block when fewer than bb rows are violated. Stack the selected rows in UBU_B, the offsets in βB\beta_B, and set s=UBw−βBs = U_Bw - \beta_B. The block problem retains these signed residuals and minimizes displacement into the halfspace intersection:

min⁡Δ12∥Δ∥2s.t.UBΔ+s≤0.(3)\min_{\Delta} \frac{1}{2}\|\Delta\|^2 \quad\text{s.t.}\quad U_B\Delta+ s \le0. \tag*{(3)}

Containment makes this quadratic program (QP) feasible. For b=1b = 1, Eq. (3) gives the projection step from ww onto the selected halfspace.

Dual PGD

Abbreviate U=UBU = U_B and GB=UU⊤G_B = UU^\top. The dual objective and updates are

max⁡λ≥0ϕ(λ)=s⊤λ−12∥U⊤λ∥2,\max_{\lambda\ge0} \phi(\lambda) = s^\top\lambda- \frac{1}{2}\|U^\top\lambda\|^2,
λj+1=[λj+s−GBλjLB]+,Δ=−U⊤λT.(4)\lambda^{j+1} = \left[\lambda^j + \frac{s-G_B\lambda^j}{L_B}\right]_+,\qquad\Delta= -U^\top\lambda^T. \tag*{(4)}

Each block takes T≥1T \ge1 steps from λ0=0\lambda^0 = 0, reusing GBG_B. The dual variables weight row normals to give the displacement. We take λmax⁡(GB)≤LB≤Lˉ<∞\lambda_{\max}(G_B) \le L_B \le\bar{L} < \infty, with coefficients and steps fixed within each block; zero blocks stay still. Normalization permits Lˉ=b\bar{L} = b; Appendix A.4 covers broader steps.

Given residuals and (sub)gradients, block updates use matrix products, reductions, and elementwise clipping, without per-block factorizations or linear solves. Fixed block shapes support GPU batching; complexity is in Appendix B.2.

Boundary fitting targets hyperplanes ([26, 6]), while violation descent minimizes the squared positive residual ([12]). Our block problem minimizes displacement under halfspace inequalities.

Learnable mobility

We introduce a positive-definite mobility PP to control directional displacement costs. Joint training learns P=Pψ(x)P = P_\psi(x) together with the predictor fθf_\theta, improving the initial prediction and subsequent repair geometry. Postprocessing prescribes PP; P=InP = I_n recovers Euclidean repair.

We use P=diag⁡(pj)≻0P = \operatorname{diag}(p_j) \succ0 and Pz=N⊤PN≻0P_z = N^\top P N \succ0, fixed during one complete repair. Let ABA_B stack selected unnormalized rows (aik)⊤(a_i^k)^\top and sBraw=(ℓik(w))i∈Bs_B^{\mathrm{raw}} = (\ell_i^k(w))_{i\in B} contain their signed linearized residuals. We replace Eq. (3) with

min⁡Δz12Δz⊤Pz−1Δzs.t.ABΔz+sBraw≤0.(5)\min_{\Delta z} \frac{1}{2}\Delta z^\top P_z^{-1}\Delta z \quad\text{s.t.}\quad A_B\Delta z+s_B^{\mathrm{raw}}\le0. \tag*{(5)}

Larger mobility eigenvalues lower the cost of the corresponding directions. Selection uses [ℓik(w)]+/(aik)⊤Pzaik[\ell_i^k(w)]_+/\sqrt{(a_i^k)^\top P_z a_i^k}, and the feasible set is unchanged. The rectangular factor R=N⊤P1/2∈Rq×nR = N^\top P^{1/2} \in\mathbb{R}^{q\times n} satisfies RR⊤=PzR R^\top= P_z. Setting z=Rνz = R\nu, with ν∈Rn\nu\in\mathbb{R}^n, reuses the Euclidean solver (Algorithm 1). Parameterization and training are in Appendix B; finite-step equivalence is in Appendix A.1.

Algorithm 1 Convex-SKM-Net repair with prescribed mobility

Require: xx, y^\hat{y}, mobility P=diag⁡(p)≻0P = \operatorname{diag}(p) \succ0, block size bb, outer cap KK, inner steps S≥1S \ge1, dual depth T≥1T \ge1, tolerance ε\varepsilon

References

  1. [1]Akshay Agrawal, Brandon Amos, Shane Barratt, Stephen Boyd, Steven Diamond, and J. Zico Kolter. Differentiable convex optimization layers. In Advances in Neural Information Processing Systems, volume 32, 2019. URL https://proceedings.neurips.cc/paper/2019/hash/9ce3c52f c432e2053399d3181c638–Abstract.html.arxiv.org/abs/1910.12430
  2. [2]Mehrazin Alizadeh and Hina Tabassum. Power control with QoS guarantees: A differentiable projection-based unsupervised learning framework. IEEE Transactions on Communications, 71 (8):4605–4619, 2023. doi: 10.1109/TCOMM.2023.3282220. URL https://doi.org/10.1109/TCOMM.2023.3282220.
  3. [3]Brandon Amos and J. Zico Kolter. OptNet: Differentiable optimization as a layer in neural networks. In Proceedings of the 34th International Conference on Machine Learning, volume 70 of Proceedings of Machine Learning Research, pp. 136–145. PMLR, 2017. URL https://proceedings.mlr.press/v70/amos17a.html.arxiv.org/abs/1703.00443
  4. [4]Randall Balestriero and Yann LeCun. POLICE: Provably optimal linear constraint enforcement for deep neural networks. In 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 1–5, 2023. doi: 10.1109/ICASSP49357.2023.10096520. URL https://doi.org/10.1109/ICASSP49357.2023.10096520.
  5. [5]Heinz H. Bauschke and Patrick L. Combettes. Convex Analysis and Monotone Operator Theory in Hilbert Spaces. Springer, second edition, 2017. doi: 10.1007/978-3-319-48311-5.DOI
  6. [6]Jonathan Briskman and Deanna Needell. Block Kaczmarz method with inequalities. Journal of Mathematical Imaging and Vision, 52(3):385–396, 2015. doi: 10.1007/s10851-014-0539-7.DOI
  7. [7]Steven Chen, Kelsey Saulnier, Nikolay Atanasov, Daniel D. Lee, Vijay Kumar, George J. Pappas, and Manfred Morari. Approximating explicit model predictive control using constrained neural networks. In 2018 Annual American Control Conference, pp. 1520–1527, 2018. doi: 10.23919/ACC.2018.8431275.
  8. [8]Ya-Chi Chu, Alkiviades Boukas, and Madeleine Udell. SnareNet: Flexible repair layers for neural networks with hard constraints. arXiv preprint arXiv:2602.09317, 2026. URL https://arxiv.org/abs/2602.09317.
  9. [9]Gonzalo E. Constante, Hao Chen, and Can Li. Enforcing hard linear constraints in deep learning models with decision rules. In Advances in Neural Information Processing Systems, volume 38, pp. 105075–105100, 2025. doi: 10.52202/ 085713-3510. URL https://papers.nips.cc/paper_files/paper/2025/hash/975a1a8e86e72c148aa2152243953064-Abstract-Conference.html.arxiv.org/abs/2505.13858
  10. [10]Rares Cristian, Pavithra Harsha, Georgia Perakis, Brian L. Quanz, and Ioannis Spantidakis. End-to-end learning for optimization via constraint-enforcing approximators. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pp. 7253–7260, 2023. doi: 10.1609/aaai.v37i6.25884.DOI
  11. [11]Jesús A. De Loera, Jamie Haddock, and Deanna Needell. A sampling Kaczmarz–Motzkin algorithm for linear feasibility. SIAM Journal on Scientific Computing, 39(5):S66–S87, 2017. doi: 10.1137/16M1073807.DOI
  12. [12]Priya L. Donti, David Rolnick, and J. Zico Kolter. DC3: A learning method for optimization with hard constraints. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=V1ZHVxJ6dSS.arxiv.org/abs/2104.12225
  13. [13]Lucio T. dos Santos. A parallel subgradient projections method for the convex feasibility problem. Journal of Computational and Applied Mathematics, 18(3):307–320, 1987. doi: 10.1016/0377-0427(87)90004-5.
  14. [14]Thomas Frerix, Matthias Niessner, and Daniel Cremers. Homogeneous linear inequality constraints for neural network activations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp. 748–749, 2020. URL https://openaccess.thecvf.com/content_CVPRW_2020/html/w45/Frerix_Homogeneous_Linear_Inequality_Constraints_for_Neural_Network_Activations_CVPRW_2020_paper.html.arxiv.org/abs/1902.01785
  15. [15]Andrea Goertzen, Kaveh Alim, Youngjae Min, and Navid Azizan. HardNet++: Nonlinear constraint enforcement in neural networks. IEEE Control Systems Letters, 10:1951–1956, 2026. doi: 10.1109/LCSYS.2026.3707460. URL https://doi.org/10.1109/LCSYS.2026.3707460.
  16. [16]Panagiotis D. Grontas, Antonio Terpin, Efe C. Balta, Raffaello D’Andrea, and John Lygeros. INet: Optimizing hard-constrained neural networks with orthogonal projection layers. In International Conference on Learning Representations, 2026. URL https://arxiv.org/abs/2508.10480.
  17. [17]Ashfaq Iftakher, Rahul Golder, Bimol Nath Roy, and M. M. Faruque Hasan. Physics-informed neural networks with hard nonlinear equality and inequality constraints. Computers & Chemical Engineering, 204:109418, 2026. doi: 10.1016/j.compchemeng.2025.109418.
  18. [18]Krzysztof C. Kiwiel. Block-iterative surrogate projection methods for convex feasibility problems. Linear Algebra and its Applications, 215:225–259, 1995. doi: 10.1016/0024-3795(93)00089-I.DOI
  19. [19]Andrei V. Konstantinov and Lev V. Utkin. A new computationally simple approach for implementing neural networks with output hard constraints. Doklady Mathematics, 108(S2):S233–S241, 2023. doi: 10.1134/S1064562423701077. URL https://doi.org/10.1134/S1064562423701077.
  20. [20]Giacomo Lastrucci and Artur M. Schweidtmann. ENFORCE: Nonlinear constrained learning with adaptive-depth neural projection. arXiv preprint arXiv:2502.06774, 2025. URL https://arxiv.org/abs/2502.06774v3.
  21. [21]Meiyi Li, Soheil Kolouri, and Javad Mohammadi. Learning to solve optimization problems with hard linear constraints. IEEE Access, 11:59995–60004, 2023. doi: 10.1109/ACCESS.2023.3285199.DOI
  22. [22]Enming Liang and Minghua Chen. Efficient bisection projection to ensure neural-network solution feasibility for optimization over general set. In Proceedings of the 42nd International Conference on Machine Learning, volume 267 of Proceedings of Machine Learning Research, pp. 37071–37099. PMLR, 2025. URL https://proceedings.mlr.press/v267/liang25b.html.
  23. [23]Emming Liang, Minghua Chen, and Steven H. Low. Homeomorphic projection to ensure neural-network solution feasibility for constrained optimization. Journal of Machine Learning Research, 25(329):1–55, 2024. URL https://jmlr.org/papers/v25/23-1577.html.
  24. [24]Youngjae Min and Navid Azizan. HardNet: Hard-constrained neural networks with universal approximation guarantees. arXiv preprint arXiv:2410.10807, 2025. URL https://arxiv.org/abs/2410.10807v4.
  25. [25]Ion Necoara and Angelia Nedić. Minibatch stochastic subgradient-based projection algorithms for feasibility problems with convex inequalities. Computational Optimization and Applications, 80(1):121–152, 2021. doi: 10.1007/s10589-021-00294-3. URL https://doi.org/10.1007/s10589-021-00294-3.
  26. [26]Deanna Needell and Joel A. Tropp. Paved with good intentions: Analysis of a randomized block Kaczmarz method. Linear Algebra and its Applications, 441:199–221, 2014. doi: 10.1016/j.laa.2012.12.022.DOI
  27. [27]Hoang T. Nguyen and Priya L. Donti. FSNet: Feasibility-seeking neural network for constrained optimization with guarantees. In Advances in Neural Information Processing Systems, volume 38, pp. 39670–39708, 2025. URL https://proceedings.neurips.cc/paper_files/paper/2025/hash/3874e2be479a9d4e94d4514046c1f934-Abstract-Conference.html.arxiv.org/abs/2506.00362
  28. [28]Philipp J. Schneider and Daniel Kuhn. Soft-radial projection for constrained end-to-end learning. arXiv preprint arXiv:2602.03461, 2026. URL https://arxiv.org/abs/2602.03461.
  29. [29]Mihaela C. Stoian, Salijona Dyrmishi, Maxime Cordy, Thomas Lukasi ewicz, and Eleonora Giunchiglia. How realistic is your synthetic data? constraining deep generative models for tabular data. In International Conference on Learning Representations, pp. 32167–32204, 2024. URL https://proceedings.iclr.cc/paper_files/paper/2024/hash/887932131fddf943e8fe3310b62c0147-Abstract-Conference.html.
  30. [30]Sunbochen Tang, Andrea Goertzen, and Navid Azizan. LMI-Net: Linear matrix inequality-constrained neural networks via differentiable projection layers. arXiv preprint arXiv:2604.05374, 2026. URL https://arxiv.org/abs/2604.05374.
  31. [31]Jesus Tordesillas, Victor Klemm, Jonathan P. How, and Marco Hutter. RAYEN: Imposition of hard convex constraints on neural networks. arXiv preprint arXiv:2307.08336v2, 2026. URL https://arxiv.org/abs/2307.08336v2.
  32. [32]Runzhong Wang, Yunhao Zhang, Xiao Guo, Tianyi Chen, Xiaokang Yang, and Junchi Yan. LinSATNet: The positive linear satisfiability neural networks. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pp. 36605–36625. PMLR, 2023. URL https://proceedings.mlr.press/v202/wang23at.html.
  33. [33]Hongtai Zeng, Chao Yang, Yanzhen Zhou, Cheng Yang, and Qinglai Guo. GLinSAT: The general linear satisfiability neural network layer by accelerated gradient descent. In Advances in Neural Information Processing Systems, volume 37, pp. 122584–122615, 2024. doi: 10.52202/079017-3896.DOI
  34. [34]Feiyu Zhang, Wendi Bao, Weiguo Li, and Qin Wang. On sampling Kaczmarz–Motzkin methods for solving large-scale nonlinear systems. Computational and Applied Mathematics, 42(3):126, 2023. doi: 10.1007/s40314-023-02265-2. URL https://doi.org/10.1007/s40314-023-02265-2.
  35. [35]Xiaochen Zhang, Haoyu Zhu, Yao Zhang, and Qingchun Hou. AT-SKM-Net: An accelerated trainable sampling kaczmarz-motzkin framework for linear hard-constraint feasibility on dynamic graphs. arXiv preprint arXiv:2609.30088, 2026. URL https://arxiv.org/abs/2609.30088.
  36. [36]Yanjun Zhang and Hanyu Li. Block sampling Kaczmarz–Motzkin methods for consistent linear systems. Calcolo, 58(3):39, 2021. doi: 10.1007/s10092-021-00429-2.DOI

Paper details

vlnc:W_pnfuahhu
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

generative AI tools

View statement
Looking for related papers…
Contents