Pith. sign in

REVIEW 4 major objections 5 minor 123 references

Deep Loss Convexification for Learning Iterative Models

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 claims that training a deep iterative model to make its test-time loss landscape star-convex around each ground truth yields provably near-optimal predictions, without changing the network architecture.

desk verdict A credible extension of the authors' PRISE idea to iterative networks and point cloud registration, with a theoretical guarantee that is honest but unverified. read the letter →

arxiv 2411.10649 v1 pith:OPMPV2JO submitted 2024-11-16 cs.CV

Ziming Zhang , Yuping Shao , Yiqing Zhang , Fangzhou Lin , Haichong Zhang , Elke Rundensteiner This is my paper · ORCID
classification cs.CV
keywords deeplossconvexificationstar-convexityiterativemodelslandscapereshapingpointcloudregistrationhomographyestimationrecurrentneuralnetworkshinge
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

Deep Loss Convexification (DLC) is a training strategy for iterative models that targets their tendency to stall at bad local optima such as saddle points. Instead of studying the loss surface with respect to network weights, DLC reshapes the loss surface with respect to the prediction variable so that, around each ground truth, it is locally star-convex, meaning unimodal along every line through the minimum. The paper claims this reshaping, enforced by three hinge losses appended to the original training loss, makes gradient-based iterations at test time converge to near-optimal predictions, with a prediction-error bound of ||omega - omega*|| <= 2L/mu for an L-Lipschitz, mu-strongly star-convex learned loss. The practical payoff is that existing architectures such as LSTM, DCP, PRNet, and DeepLK can be improved on recurrent classification, 3D point cloud registration, and multimodal image alignment without any change to inference-time computation.

What carries the argument

The load-bearing object is star-convexity, a structured nonconvexity in which a function is unimodal on all lines passing through a global minimizer, so a downhill path to the optimum is visible from every point. Strong star-convexity adds a quadratic curvature parameter mu, and Lemma 1 converts that geometric picture into two inequality constraints that involve no gradients, making them usable as loss terms. The machinery is the trio of hinge constraints from Eqs. 12-14 paired with a sampling-based training loop (Algorithm 1) that generates noisy neighbors omega_i of each ground truth and enforces the inequalities at those sampled points. This deep reparametrization, where the network's prediction variable replaces the original loss parameter, is what enables the learned loss landscape to be locally convex-like around each ground truth.

What would settle it

Train DLC on any of the reported tasks, then on held-out data sample neighborhoods around ground truths and test the three inequalities (Eqs. 12-14) directly, estimating the Lipschitz constant L and strong-convexity constant mu numerically; if a substantial fraction of samples violate the inequalities, or mu comes out near zero, then Lemma 2's bound ||omega - omega*|| <= 2L/mu is vacuous and DLC's accuracy gains need a different explanation.

Watch

Extended reading notes

Core claim

The paper's central claim is that an overparameterized network can be trained not merely to map inputs to predictions but to sculpt the geometry of its own loss around those predictions. Concretely, DLC adds three contrastive hinge losses to the training objective, each enforcing one facet of strong star-convexity: the ground truth is a local minimum (Eq. 12), a quadratic lower envelope anchors the loss at the optimum (Eq. 13), and every chord between the optimum and any other point lies above the function value (Eq. 14). If the learned loss h_{theta*} is L-Lipschitz and mu-strongly star-convex, Lemma 2 bounds the distance from any test-time prediction to the ground truth by 2L/mu, and Lemma 3 gives an O(1/T) average-error reduction when predictions are averaged over iterations. The paper argues that the training objective makes these conditions hold approximately in a neighborhood of every ground truth, so iterative inference inherits the convergence guarantees of star-convex optimization.

Load-bearing premise

The guarantee rests on the assumption that the trained network actually produces a loss landscape that is star-convex, L-Lipschitz, and mu-strongly star-convex around each ground truth; training only enforces the three inequalities at finitely many randomly sampled neighbors with slack variables, so nothing certifies the learned landscape satisfies the assumed geometry.

Editorial extensions

If this is right

  • Under the learned star-convex landscape, every test-time gradient iteration keeps the prediction within a ball of radius 2L/mu around the ground truth, so iterative refinement cannot wander far from the optimum.
  • Averaging predictions across iterations reduces the expected squared error at a rate O(1/T) (Lemma 3), so longer inference-time iterations are predictable and safe.
  • Because DLC changes only the training objective, it can be layered onto existing architectures without changing inference graphs or adding test-time cost.
  • Across the three task families studied, DLC improves registration and alignment accuracy over the same backbones, including transfer to unseen classes from ModelNet40 to ShapeNetCore and after ICP refinement.

Reading between the lines

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

  • Because the proof's guarantee holds only when the learned landscape is truly star-convex, and training merely pushes sampled inequalities, a practical certificate would require checking the inequalities on dense neighborhoods; absent that, the theoretical bound is an aspiration rather than a verified property.
  • Lemma 3's averaging benefit relies on uncorrelated per-iteration errors, an assumption the authors admit is 'very strong' and not satisfied in their setting, so DLC's empirical gains may stem from landscape smoothing rather than from this specific averaging bound.
  • The same star-convexification recipe could apply to other test-time iterative refinement loops, such as pose optimization, optical flow, or generative sampling, wherever a differentiable loss is optimized at inference.
  • A comparison with the authors' prior PRISE work suggests that dropping the adversarial max over hinge losses slightly improves training stability, indicating the convexification effect is robust to how the constraints are aggregated.
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 Deep Loss Convexification (DLC), a training-time regularizer that appends three hinge losses based on local star-convexity constraints on the loss landscape with respect to test-time predictions omega, rather than with respect to network weights. The stated goal is to reshape the loss landscape so that iterative inference by gradient descent converges to near-optimal solutions. The method is evaluated on three tasks: Pixel-MNIST classification with LSTM, 3D point cloud registration with DCP and PRNet, and multimodal image alignment with DeepLK-based architecture. The authors report state-of-the-art results on point cloud registration, improved LSTM accuracy, and image alignment numbers that match their prior PRISE work.

Significance. The core idea of shaping the loss landscape in prediction space via star-convexity constraints is novel and potentially impactful, as it is architecture-agnostic and adds no inference-time cost. The formulation of star-convexity conditions as differentiable hinge losses is clean and easy to plug into existing iterative models. The point-cloud rotation improvements are substantial and consistent across datasets, and the transferability and scalability experiments add useful evidence. However, the claimed near-optimality guarantee is conditional on an unverified perfect-learning assumption, and the empirical support is mixed: translation errors on ModelNet40 are sometimes worse than the baselines, and the image-alignment experiments do not provide new numbers beyond the authors' prior PRISE paper. These issues materially affect the strength of the central claims, but the underlying idea remains viable and is worth a major revision.

major comments (4)
  1. [Sec. 3.2.3, Lemma 2, Eqs. (11)-(14)] The near-optimality bound in Eq. (15) is derived under the assumption that the network can learn star-convex loss landscapes w.r.t. predictions perfectly, but the training objective in Eqs. (11)-(14) only penalizes the three star-convexity inequalities at finitely many randomly sampled neighbors omega_i, with slack variables, and there is no verification that the learned h_{theta*} satisfies Eq. (13) for held-out inputs or for the actual test-time iterates, nor any measurement of L or mu. As a consequence, Eq. (15) is not an established guarantee for the trained model; it holds only under an unverified strong assumption. The paper should either provide a finite-sample or generalization bound connecting satisfaction of the sampled constraints to the required landscape property, or empirically measure the landscape constants and violation rates on held-out data, or explicitly restate the contribution as an empirical regularizer without the near-optimality guarantee.
  2. [Sec. 3.2.3, Lemma 3] The O(1/T) error bound in Eq. (17) depends on the assumption that prediction errors across iterations are uncorrelated, and the paper itself acknowledges that this assumption 'seems not to hold' (Sec. 3.2.3). Since the averaging scheme in Eq. (10) and the tuned test-time iterations in Sec. 4.2 are presented as part of the method's benefit, this result should not be stated as a theoretical justification; at minimum it needs a clearly labeled heuristic status or a correlation-robust analysis.
  3. [Table 2, ModelNet40 rows] The translation results contradict the claim of consistent improvement: for DCP, DLC+DCP has higher MSE(T) than DCP both without ICP refinement (2.55e-5 vs. 1.79e-5) and with ICP refinement (1.63e-6 vs. 2.72e-7), and for PRNet it is worse without ICP refinement (0.0002 vs. 0.0001). The text in Sec. 4.2.3 says DLC 'sometimes has little help' in translation, and Sec. 4.2.4 says the approach 'almost consistently and significantly' improves performance; these statements need to be qualified to reflect the actual direction and magnitude of the translation results, ideally with confidence intervals.
  4. [Sec. 4.3, Table 5] The image-alignment experiments do not report DLC's own numbers; Table 5 lists PRISE [20], and the text states the results are 'almost identical' to [20]. Since the abstract claims state-of-the-art performance with DLC on this task, please provide the actual DLC numbers with standard deviations and a statistical comparison against PRISE, or remove the state-of-the-art claim for this task.
minor comments (5)
  1. [Title and Abstract] The paper consistently uses 'multimodel image alignment' where 'multimodal image alignment' appears intended; please correct the typo.
  2. [Sec. 4.2.1, Figure 4] The loss landscape slices are two-dimensional projections, so statements about 'convex-like' shapes are only qualitative; please add a quantitative metric or at least a cautionary caption note.
  3. [Table 3, 'Unseen Class 14' row] The value 8.0e-8 for MSE(R) with ICP refinement is an outlier compared to the other entries; please verify whether this is a typo or a reporting artifact.
  4. [Sec. 3.1, Lemma 1 proof] The statement 'since nabla f(omega*) = 0' assumes differentiability at the global minimum; please state the required regularity condition explicitly.
  5. [General] The paper does not mention code or data release; providing an implementation would strengthen reproducibility, especially for the sampling strategy in Algorithm 1.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: Lemma 2 is a conditional consequence of the strong star-convexity constraint under an explicit, unverified assumption, and the only self-citation is disclosed and not load-bearing.

full rationale

The paper's central guarantee is conditional rather than circular. Lemma 2 (Sec. 3.2.3) states that if h_theta* is L-Lipschitz and mu-strongly star-convex, then ||omega - omega*|| <= 2L/mu, and its proof directly uses Eq. 13 plus the Lipschitz assumption. This is a standard consequence of the strong star-convexity inequality, not a fitted quantity renamed as a prediction; the bound is not used to construct the training objective, and the paper explicitly hedges the premise with 'assuming that the networks can learn star-convex loss landscapes w.r.t. predictions perfectly' (Sec. 3.2.3). The training objective (Eq. 11) only soft-penalizes violations at finitely sampled neighbors with slack variables, so the learned landscape may not actually satisfy the assumption for held-out data or test-time iterates. That gap is an unverified assumption and a correctness risk, not circularity, because the theorem is stated under an explicit hypothesis and does not redefine the prediction as the constraint. Lemma 3 is likewise explicitly hedged: 'the uncorrelated assumption is very strong and in our case it seems not to hold' (Sec. 3.2.3), again a validity caveat rather than a circular reduction. The only self-citation signal is the disclosed dependence on prior work [20] (PRISE) for the image-alignment evaluation and as the predecessor framework, noted in Sec. 1 ('This work is a significant extension of [20]') and Sec. 4.3. That citation is not load-bearing for the central guarantee, which rests on external star-convexity results [16], [17]. The experiments are evaluated against external benchmarks and existing baselines (Pixel-MNIST leaderboard, ModelNet40, ShapeNetCore, 3DMatch, KITTI, MSCOCO, GoogleEarth, GoogleMap) rather than on constants fitted to the test set, so the empirical contribution is self-contained. Overall, the derivation is self-contained as a conditional analysis; the main weakness is that the learned loss landscape is never certified to be star-convex, L-Lipschitz, or mu-strongly star-convex, which should be treated as a correctness risk rather than circularity.

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

The central theoretical result rests on unverified realizability of star-convex landscapes plus a handful of tuned hyperparameters. No new physical or mathematical entities are introduced. The main free parameters are the three star-convexity weights and the sampling/iteration counts, all chosen by grid search or trainable reparametrization.

free parameters (5)
  • rho = 0.6 for DLC+DCP ModelNet40, 1.0 for DLC+PRNet ModelNet40 and 3DMatch
    Trade-off weight between original loss and star-convexity hinge losses, tuned by grid search per dataset and task.
  • lambda = 0.5 by default; reparametrized as trainable sigmoid in some experiments
    Interpolation coefficient between ground truth and noisy samples in star-convexity constraints; treated as a hyperparameter.
  • mu = 4.0 by default in image alignment; reparametrized as exponential trainable parameter in point cloud setting
    Surface sharpness constant in strong star-convexity inequalities; controls the quadratic envelope and is chosen by grid search.
  • number of noisy samples per ground truth = 3 by default
    Number of random omega samples used to approximate the expected hinge losses in Eq. 11; selected by grid search.
  • maximum test-time iterations T = Tuned per dataset, up to 5; best result reported
    Test-time iteration count is selected on the test set and the best performance is reported, which can inflate apparent gains.
assumptions (5)
  • domain assumption If the trained loss landscape is L-Lipschitz and mu-strongly star-convex, then Lemma 2 bounds prediction error by 2L/mu.
    Section 3.2.3 explicitly assumes the network learns star-convex loss landscapes perfectly. This is not verified empirically.
  • ad hoc to paper The three hinge-loss constraints enforced at finitely many random samples are sufficient to make the loss landscape approximately star-convex.
    Algorithm 1 samples only a fixed number of neighbors per ground truth; the paper provides no guarantee that this enforces the continuous inequalities.
  • domain assumption Test-time fixed-point iteration can be analyzed using gradient-descent convergence results for star-convex functions.
    Section 3.2.1 and 3.2.3 rely on GD guarantees from [13], [17], but inference uses average fixed-point iteration rather than explicit gradient descent on the loss.
  • domain assumption Prediction errors across iterations are uncorrelated.
    Lemma 3 relies on this assumption; the authors state it is 'very strong and in our case it seems not to hold.'
  • domain assumption Overparametrization gives neural networks the capacity to reshape loss landscapes as desired.
    The method is motivated by overparametrization in Section 1 and Section 3.2.1, but no capacity or realizability result is proved.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Loss Convexification for Learning Iterative Models." pith.science (2026). https://pith.science/paper/OPMPV2JO

@misc{pith2026241110649,
  author       = {Pith},
  title        = {Pith review of: Deep Loss Convexification for Learning Iterative Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OPMPV2JO}},
  note         = {Machine review of arXiv:2411.10649}
}
read the original abstract

Iterative methods such as iterative closest point (ICP) for point cloud registration often suffer from bad local optimality (e.g. saddle points), due to the nature of nonconvex optimization. To address this fundamental challenge, in this paper we propose learning to form the loss landscape of a deep iterative method w.r.t. predictions at test time into a convex-like shape locally around each ground truth given data, namely Deep Loss Convexification (DLC), thanks to the overparametrization in neural networks. To this end, we formulate our learning objective based on adversarial training by manipulating the ground-truth predictions, rather than input data. In particular, we propose using star-convexity, a family of structured nonconvex functions that are unimodal on all lines that pass through a global minimizer, as our geometric constraint for reshaping loss landscapes, leading to (1) extra novel hinge losses appended to the original loss and (2) near-optimal predictions. We demonstrate the state-of-the-art performance using DLC with existing network architectures for the tasks of training recurrent neural networks (RNNs), 3D point cloud registration, and multimodel image alignment.

Figures

Figures reproduced from arXiv: 2411.10649 by the authors.

Figure 1
Figure 1. Illustration of iterative point cloud registration with the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Illustration of differences in loss landscapes [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Performance comparison in both training and testing for [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Test-time loss landscape comparison using PRNet on ModelNet40: [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Illustration of test-time point cloud registration for [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Illustration of test-time performance comparison with [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Comparison results for PRNet vs. DLC+PRNet. consistently improve the LSTM performance in both cases. We further list our comparison results in [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Training loss comparison at stage 3 on Google Earth. Conditions 1-3 correspond to the constraints in Eqs. 12-14 in order. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Performance comparison on Google Earth using the [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Pixel error vs. various hyperparameters in Eqs. 12-14 on Google Earth. illustrate our performance comparison in [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

123 extracted references · 56 canonical work pages

  1. [20]

    Prise: Demystifying deep lucas- kanade with strongly star-convex constraints for multimodel image alignment,

    Y . Zhang, X. Huang, and Z. Zhang, “Prise: Demystifying deep lucas- kanade with strongly star-convex constraints for multimodel image alignment,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 13 187–13 197

    2023

  2. [1]

    A comprehensive survey on point cloud registration,

    X. Huang, G. Mei, J. Zhang, and R. Abbas, “A comprehensive survey on point cloud registration,” arXiv preprint arXiv:2103.02690, 2021

    arXiv 2021

  3. [2]

    Least-squares fitting of two 3-d point sets,

    K. S. Arun, T. S. Huang, and S. D. Blostein, “Least-squares fitting of two 3-d point sets,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-9, no. 5, pp. 698–700, 1987

    1987

  4. [3]

    Prnet: Self-supervised learning for partial- to-partial registration,

    Y . Wang and J. M. Solomon, “Prnet: Self-supervised learning for partial- to-partial registration,” Advances in neural information processing sys- tems, vol. 32, 2019

    2019

  5. [4]

    D. R. Smart, Fixed point theorems. Cup Archive, 1980, vol. 66

    1980

  6. [5]

    A survey of optimization methods from a machine learning perspective,

    S. Sun, Z. Cao, H. Zhu, and J. Zhao, “A survey of optimization methods from a machine learning perspective,” IEEE transactions on cybernetics, vol. 50, no. 8, pp. 3668–3681, 2019

    2019

  7. [6]

    Recent theoretical advances in non- convex optimization,

    M. Danilova, P. Dvurechensky, A. Gasnikov, E. Gorbunov, S. Guminov, D. Kamzolov, and I. Shibaev, “Recent theoretical advances in non- convex optimization,” arXiv preprint arXiv:2012.06188, 2020

    arXiv 2012

  8. [7]

    The power of convex relaxation: Near-optimal matrix completion,

    E. J. Candès and T. Tao, “The power of convex relaxation: Near-optimal matrix completion,” IEEE Transactions on Information Theory, vol. 56, no. 5, pp. 2053–2080, 2010

    2010

Show all 123 references
  1. [8]

    Non-convex optimization for machine learning,

    P. Jain, P. Kar et al., “Non-convex optimization for machine learning,” Foundations and Trends® in Machine Learning , vol. 10, no. 3-4, pp. 142–363, 2017

    2017

  2. [9]

    Exact matrix completion via convex opti- mization,

    E. Candes and B. Recht, “Exact matrix completion via convex opti- mization,” Communications of the ACM , vol. 55, no. 6, pp. 111–119, 2012

    2012

  3. [10]

    An alternative view: When does sgd escape local minima?

    B. Kleinberg, Y . Li, and Y . Yuan, “An alternative view: When does sgd escape local minima?” in International Conference on Machine Learning. PMLR, 2018, pp. 2698–2707

    2018

  4. [11]

    Visualizing the loss landscape of neural nets,

    H. Li, Z. Xu, G. Taylor, C. Studer, and T. Goldstein, “Visualizing the loss landscape of neural nets,” Advances in neural information processing systems, vol. 31, 2018

    2018

  5. [12]

    SGD converges to global minimum in deep learning via star-convex path,

    Y . Zhou, J. Yang, H. Zhang, Y . Liang, and V . Tarokh, “SGD converges to global minimum in deep learning via star-convex path,” inInternational Conference on Learning Representations, 2019

    2019

  6. [13]

    Near-optimal methods for min- imizing star-convex functions and beyond,

    O. Hinder, A. Sidford, and N. Sohoni, “Near-optimal methods for min- imizing star-convex functions and beyond,” in Conference on learning theory. PMLR, 2020, pp. 1894–1938

    2020

  7. [14]

    To- wards understanding the role of over-parametrization in generalization of neural networks,

    B. Neyshabur, Z. Li, S. Bhojanapalli, Y . LeCun, and N. Srebro, “To- wards understanding the role of over-parametrization in generalization of neural networks,” arXiv preprint arXiv:1805.12076, 2018

    2018 arXiv

  8. [15]

    Cubic regularization of newton method and its global performance,

    Y . Nesterov and B. T. Polyak, “Cubic regularization of newton method and its global performance,” Mathematical Programming , vol. 108, no. 1, pp. 177–205, 2006

    2006

  9. [16]

    Optimizing star-convex functions,

    J. C. Lee and P. Valiant, “Optimizing star-convex functions,” in 2016 IEEE 57th Annual Symposium on Foundations of Computer Science (FOCS). IEEE, 2016, pp. 603–614

    2016

  10. [17]

    Near-optimal methods for minimizing star-convex functions and beyond,

    O. Hinder, A. Sidford, and N. Sohoni, “Near-optimal methods for minimizing star-convex functions and beyond,” inProceedings of Thirty Third Conference on Learning Theory, 2020, pp. 1894–1938

    2020

  11. [18]

    Sgd for structured nonconvex functions: Learning rates, minibatching and interpolation,

    R. Gower, O. Sebbouh, and N. Loizou, “Sgd for structured nonconvex functions: Learning rates, minibatching and interpolation,” in Interna- tional Conference on Artificial Intelligence and Statistics . PMLR, 2021, pp. 1315–1323

    2021

  12. [19]

    Sequential subspace optimization for quasar-convex optimization problems with inexact gradient,

    I. A. Kuruzov and F. S. Stonyakin, “Sequential subspace optimization for quasar-convex optimization problems with inexact gradient,” in International Conference on Optimization and Applications. Springer, 2021, pp. 19–33

    2021

  13. [21]

    Adversarial weight perturbation helps robust generalization,

    D. Wu, S.-T. Xia, and Y . Wang, “Adversarial weight perturbation helps robust generalization,” in NeurIPS, 2020

    2020

  14. [22]

    Lossplot: A better way to visualize loss landscapes,

    R. Bain, M. Tokarev, H. Kothari, and R. Damineni, “Lossplot: A better way to visualize loss landscapes,” arXiv preprint arXiv:2111.15133 , 2021. IEEE TRANSACTION ON PATTERN ANAL YSIS AND MACHINE INTELLIGENCE 10 Table 5: Performance comparison on MSCOCO, GoogleEarth, and Google...

    2021 arXiv

  15. [23]

    Deep ensembles: A loss landscape perspective,

    S. Fort, H. Hu, and B. Lakshminarayanan, “Deep ensembles: A loss landscape perspective,” arXiv preprint arXiv:1912.02757, 2019

    1912 arXiv

  16. [24]

    Exploring the landscape of spatial robustness,

    L. Engstrom, B. Tran, D. Tsipras, L. Schmidt, and A. Madry, “Exploring the landscape of spatial robustness,” in International conference on machine learning. PMLR, 2019, pp. 1802–1811

    2019

  17. [25]

    The loss landscape of overparameterized neural networks,

    Y . Cooper, “The loss landscape of overparameterized neural networks,” arXiv preprint arXiv:1804.10200, 2018

    2018 arXiv

  18. [26]

    Geometry of the loss landscape in overparameterized neural networks: Symmetries and invariances,

    B. Simsek, F. Ged, A. Jacot, F. Spadaro, C. Hongler, W. Gerstner, and J. Brea, “Geometry of the loss landscape in overparameterized neural networks: Symmetries and invariances,” in International Conference on Machine Learning. PMLR, 2021, pp. 9722–9732

    2021

  19. [27]

    Embedding principle of loss landscape of deep neural networks,

    Y . Zhang, Z. Zhang, T. Luo, and Z. J. Xu, “Embedding principle of loss landscape of deep neural networks,” Advances in Neural Information Processing Systems, vol. 34, pp. 14 848–14 859, 2021

    2021

  20. [28]

    The global landscape of neural networks: An overview,

    R. Sun, D. Li, S. Liang, T. Ding, and R. Srikant, “The global landscape of neural networks: An overview,” IEEE Signal Processing Magazine , vol. 37, no. 5, pp. 95–108, 2020

    2020

  21. [29]

    Low nonconvexity-rank bilinear matrix inequalities: algorithms and applications in robust controller and struc- ture designs,

    H. D. Tuan and P. Apkarian, “Low nonconvexity-rank bilinear matrix inequalities: algorithms and applications in robust controller and struc- ture designs,” IEEE Transactions on Automatic Control, vol. 45, no. 11, pp. 2111–2117, 2000

    2000

  22. [30]

    Regularized m-estimators with non- convexity: Statistical and algorithmic theory for local optima,

    P.-L. Loh and M. J. Wainwright, “Regularized m-estimators with non- convexity: Statistical and algorithmic theory for local optima,” arXiv preprint arXiv:1305.2436, 2013

    2013 arXiv

  23. [31]

    Regularized m-estimators with nonconvexity: Statistical and algorithmic theory for local optima,

    ——, “Regularized m-estimators with nonconvexity: Statistical and algorithmic theory for local optima,” The Journal of Machine Learning Research, vol. 16, no. 1, pp. 559–616, 2015

    2015

  24. [32]

    Graduated non- convexity for robust spatial perception: From non-minimal solvers to global outlier rejection,

    H. Yang, P. Antonante, V . Tzoumas, and L. Carlone, “Graduated non- convexity for robust spatial perception: From non-minimal solvers to global outlier rejection,” IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 1127–1134, 2020

    2020

  25. [33]

    Adaptively solving the local-minimum problem for deep neural networks,

    H. Wang and J. T.-H. Lo, “Adaptively solving the local-minimum problem for deep neural networks,” arXiv preprint arXiv:2012.13632 , 2020

    2012 arXiv

  26. [34]

    Successive convexification of non-convex optimal control problems and its convergence properties,

    Y . Mao, M. Szmuk, and B. Açıkme¸ se, “Successive convexification of non-convex optimal control problems and its convergence properties,” in 2016 IEEE 55th Conference on Decision and Control (CDC). IEEE, 2016, pp. 3636–3641

    2016

  27. [35]

    Adaptive meth- ods for nonconvex optimization,

    S. Reddi, M. Zaheer, D. Sachan, S. Kale, and S. Kumar, “Adaptive meth- ods for nonconvex optimization,” in Proceeding of 32nd Conference on Neural Information Processing Systems (NIPS 2018), 2018

    2018

  28. [36]

    Regularized deep learning with nonconvex penalties,

    S. Vettam and M. John, “Regularized deep learning with nonconvex penalties,” arXiv preprint arXiv:1909.05142, 2019

    1909 arXiv

  29. [37]

    Learning a similarity metric discriminatively, with application to face verification,

    S. Chopra, R. Hadsell, and Y . LeCun, “Learning a similarity metric discriminatively, with application to face verification,” in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recog- nition (CVPR’05), vol. 1. IEEE, 2005, pp. 539–546

    2005

  30. [38]

    Dimensionality reduction by learning an invariant mapping,

    R. Hadsell, S. Chopra, and Y . LeCun, “Dimensionality reduction by learning an invariant mapping,” in 2006 IEEE Computer Society Con- ference on Computer Vision and Pattern Recognition (CVPR’06), vol. 2. IEEE, 2006, pp. 1735–1742

    2006

  31. [39]

    Representation learning with contrastive predictive coding,

    A. v. d. Oord, Y . Li, and O. Vinyals, “Representation learning with contrastive predictive coding,” arXiv preprint arXiv:1807.03748, 2018

    2018 arXiv

  32. [40]

    Contrastive multiview coding,

    Y . Tian, D. Krishnan, and P. Isola, “Contrastive multiview coding,” in European conference on computer vision. Springer, 2020, pp. 776–794

    2020

  33. [41]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in International conference on machine learning. PMLR, 2020, pp. 1597–1607

    2020

  34. [42]

    Momentum contrast for unsupervised visual representation learning,

    K. He, H. Fan, Y . Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 9729–9738

    2020

  35. [43]

    Understanding contrastive representation learning through alignment and uniformity on the hypersphere,

    T. Wang and P. Isola, “Understanding contrastive representation learning through alignment and uniformity on the hypersphere,” in International Conference on Machine Learning. PMLR, 2020, pp. 9929–9939

    2020

  36. [44]

    Understanding the behaviour of contrastive loss,

    F. Wang and H. Liu, “Understanding the behaviour of contrastive loss,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 2495–2504

    2021

  37. [45]

    Contrastive learning inverts the data generating process,

    R. S. Zimmermann, Y . Sharma, S. Schneider, M. Bethge, and W. Bren- del, “Contrastive learning inverts the data generating process,” in International Conference on Machine Learning . PMLR, 2021, pp. 12 979–12 990

    2021

  38. [46]

    Contrastive boundary learning for point cloud segmentation,

    L. Tang, Y . Zhan, Z. Chen, B. Yu, and D. Tao, “Contrastive boundary learning for point cloud segmentation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 8489–8499

    2022

  39. [47]

    Unsupervised point cloud object co-segmentation by co-contrastive learning and mutual attention sampling,

    C.-K. Yang, Y .-Y . Chuang, and Y .-Y . Lin, “Unsupervised point cloud object co-segmentation by co-contrastive learning and mutual attention sampling,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 7335–7344

    2021

  40. [48]

    Contrastive representation learning: A framework and review,

    P. H. Le-Khac, G. Healy, and A. F. Smeaton, “Contrastive representation learning: A framework and review,” IEEE Access, vol. 8, pp. 193 907– 193 934, 2020

    2020

  41. [49]

    Omnet: Learning overlapping mask for partial-to-partial point cloud registration,

    H. Xu, S. Liu, G. Wang, G. Liu, and B. Zeng, “Omnet: Learning overlapping mask for partial-to-partial point cloud registration,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 3132–3141

    2021

  42. [50]

    Pointnetlk revisited,

    X. Li, J. K. Pontes, and S. Lucey, “Pointnetlk revisited,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recogni- tion, 2021, pp. 12 763–12 772

    2021

  43. [51]

    Geometric trans- former for fast and robust point cloud registration,

    Z. Qin, H. Yu, C. Wang, Y . Guo, Y . Peng, and K. Xu, “Geometric trans- former for fast and robust point cloud registration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 11 143–11 152. IEEE TRANSACTION ON PATTERN ANAL YSIS A...

    2022

  44. [52]

    Regtr: End-to-end point cloud correspon- dences with transformers,

    Z. J. Yew and G. H. Lee, “Regtr: End-to-end point cloud correspon- dences with transformers,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 6677–6686

    2022

  45. [53]

    Efficient sparse icp,

    P. Mavridis, A. Andreadis, and G. Papaioannou, “Efficient sparse icp,” Computer Aided Geometric Design, vol. 35, pp. 16–26, 2015

    2015

  46. [54]

    Outlier robust icp for minimizing fractional rmsd,

    J. M. Phillips, R. Liu, and C. Tomasi, “Outlier robust icp for minimizing fractional rmsd,” in Sixth International Conference on 3-D Digital Imaging and Modeling (3DIM 2007). IEEE, 2007, pp. 427–434

    2007

  47. [55]

    A new point matching algorithm for non- rigid registration,

    H. Chui and A. Rangarajan, “A new point matching algorithm for non- rigid registration,” Computer Vision and Image Understanding, vol. 89, no. 2-3, pp. 114–141, 2003

    2003

  48. [56]

    A polynomial-time solution for robust reg- istration with extreme outlier rates,

    H. Yang and L. Carlone, “A polynomial-time solution for robust reg- istration with extreme outlier rates,” arXiv preprint arXiv:1903.08588 , 2019

    1903 arXiv

  49. [57]

    Teaser: Fast and certifiable point cloud registration,

    H. Yang, J. Shi, and L. Carlone, “Teaser: Fast and certifiable point cloud registration,” IEEE Transactions on Robotics , vol. 37, no. 2, pp. 314–333, 2020

    2020

  50. [58]

    Pointnetlk: Robust & efficient point cloud registration using pointnet,

    Y . Aoki, H. Goforth, R. A. Srivatsan, and S. Lucey, “Pointnetlk: Robust & efficient point cloud registration using pointnet,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 7163–7172

    2019

  51. [59]

    Deep closest point: Learning represen- tations for point cloud registration,

    Y . Wang and J. M. Solomon, “Deep closest point: Learning represen- tations for point cloud registration,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 3523–3532

    2019

  52. [60]

    Pcrnet: Point cloud registration network using pointnet encoding,

    V . Sarode, X. Li, H. Goforth, Y . Aoki, R. A. Srivatsan, S. Lucey, and H. Choset, “Pcrnet: Point cloud registration network using pointnet encoding,” arXiv preprint arXiv:1908.07906, 2019

    1908 arXiv

  53. [61]

    Rpm-net: Robust point matching using learned features,

    Z. J. Yew and G. H. Lee, “Rpm-net: Robust point matching using learned features,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 11 824–11 833

    2020

  54. [62]

    Deep global registration,

    C. Choy, W. Dong, and V . Koltun, “Deep global registration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 2514–2523

    2020

  55. [63]

    Predator: Registration of 3d point clouds with low overlap,

    S. Huang, Z. Gojcic, M. Usvyatsov, A. Wieser, and K. Schindler, “Predator: Registration of 3d point clouds with low overlap,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2021, pp. 4267–4276

    2021

  56. [64]

    Deep learning based point cloud registra- tion: an overview,

    Z. Zhang, Y . Dai, and J. Sun, “Deep learning based point cloud registra- tion: an overview,”Virtual Reality & Intelligent Hardware, vol. 2, no. 3, pp. 222–246, 2020

    2020

  57. [65]

    Robust registration of multimodal remote sensing images based on structural similarity,

    Y . Ye, J. Shan, L. Bruzzone, and L. Shen, “Robust registration of multimodal remote sensing images based on structural similarity,”IEEE Transactions on Geoscience and Remote Sensing , vol. 55, no. 5, pp. 2941–2958, 2017

    2017

  58. [66]

    Adaptive context network for scene parsing,

    J. Fu, J. Liu, Y . Wang, Y . Li, Y . Bao, J. Tang, and H. Lu, “Adaptive context network for scene parsing,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 6748–6757

    2019

  59. [67]

    Fast and robust matching for multimodal remote sensing image registration,

    Y . Ye, L. Bruzzone, J. Shan, F. Bovolo, and Q. Zhu, “Fast and robust matching for multimodal remote sensing image registration,” IEEE Transactions on Geoscience and Remote Sensing , vol. 57, no. 11, pp. 9059–9070, 2019

    2019

  60. [68]

    Homography estimation from image pairs with hierarchical convolutional networks,

    F. Erlik Nowruzi, R. Laganiere, and N. Japkowicz, “Homography estimation from image pairs with hierarchical convolutional networks,” in Proceedings of the IEEE International Conference on Computer Vision Workshops, 2017, pp. 913–920

    2017

  61. [69]

    Unsupervised deep homography: A fast and robust homography esti- mation model,

    T. Nguyen, S. W. Chen, S. S. Shivakumar, C. J. Taylor, and V . Kumar, “Unsupervised deep homography: A fast and robust homography esti- mation model,” IEEE Robotics and Automation Letters , vol. 3, no. 3, pp. 2346–2353, 2018

    2018

  62. [70]

    Deep homography estima- tion for dynamic scenes,

    H. Le, F. Liu, S. Zhang, and A. Agarwala, “Deep homography estima- tion for dynamic scenes,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 7652–7661

    2020

  63. [71]

    Content-aware unsupervised deep homography estimation,

    J. Zhang, C. Wang, S. Liu, L. Jia, N. Ye, J. Wang, J. Zhou, and J. Sun, “Content-aware unsupervised deep homography estimation,” in European Conference on Computer Vision . Springer, 2020, pp. 653– 669

    2020

  64. [72]

    Deep image homogra- phy estimation,

    D. DeTone, T. Malisiewicz, and A. Rabinovich, “Deep image homogra- phy estimation,” arXiv preprint arXiv:1606.03798, 2016

    2016 arXiv

  65. [73]

    Clkn: Cascaded lucas- kanade networks for image alignment,

    C.-H. Chang, C.-N. Chou, and E. Y . Chang, “Clkn: Cascaded lucas- kanade networks for image alignment,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2017, pp. 2213–2221

    2017

  66. [74]

    Deep lucas-kanade homography for multimodal image alignment,

    Y . Zhao, X. Huang, and Z. Zhang, “Deep lucas-kanade homography for multimodal image alignment,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 15 950–15 959

    2021

  67. [75]

    Iterative deep homography estimation,

    S.-Y . Cao, J. Hu, Z. Sheng, and H.-L. Shen, “Iterative deep homography estimation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2022, pp. 1879–1888

    2022

  68. [76]

    A survey of planar homog- raphy estimation techniques,

    A. Agarwal, C. Jawahar, and P. Narayanan, “A survey of planar homog- raphy estimation techniques,”Centre for Visual Information Technology, Tech. Rep. IIIT/TR/2005/12, 2005

    2005

  69. [77]

    Deep learning of diffeomorphisms for optimal reparametrizations of shapes,

    E. Celledoni, H. Glöckner, J. Riseth, and A. Schmeding, “Deep learning of diffeomorphisms for optimal reparametrizations of shapes,” arXiv preprint arXiv:2207.11141, 2022

    2022 arXiv

  70. [78]

    Deep reparametrization of multi-frame super-resolution and denoising,

    G. Bhat, M. Danelljan, F. Yu, L. Van Gool, and R. Timofte, “Deep reparametrization of multi-frame super-resolution and denoising,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 2460–2470

    2021

  71. [79]

    An iterative image registration technique with an application to stereo vision,

    B. D. Lucas and T. Kanade, “An iterative image registration technique with an application to stereo vision,” in Proceedings of Imaging Under- standing Workshop, 1981, pp. 121–130

    1981

  72. [80]

    C. M. Bishop and N. M. Nasrabadi, Pattern recognition and machine learning. Springer, 2006, vol. 4, no. 4

    2006

  73. [81]

    Convergence analysis of two-layer neural networks with relu activation,

    Y . Li and Y . Yuan, “Convergence analysis of two-layer neural networks with relu activation,” Advances in Neural Information Processing Sys- tems, vol. 30, pp. 597–607, 2017

    2017

  74. [82]

    Smpconv: Self-moving point representations for continuous convolution,

    S. Kim and E. Park, “Smpconv: Self-moving point representations for continuous convolution,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 10 289–10 299

    2023

  75. [83]

    Efficiently modeling long sequences with structured state spaces,

    A. Gu, K. Goel, and C. Ré, “Efficiently modeling long sequences with structured state spaces,” arXiv preprint arXiv:2111.00396, 2021

    2021 arXiv

  76. [84]

    Flexconv: Continuous kernel convolutions with differentiable kernel sizes,

    D. W. Romero, R.-J. Bruintjes, J. M. Tomczak, E. J. Bekkers, M. Hoogendoorn, and J. C. van Gemert, “Flexconv: Continuous kernel convolutions with differentiable kernel sizes,” arXiv preprint arXiv:2110.08059, 2021

    2021 arXiv

  77. [85]

    Combining recurrent, convolutional, and continuous-time models with linear state space layers,

    A. Gu, I. Johnson, K. Goel, K. Saab, T. Dao, A. Rudra, and C. Ré, “Combining recurrent, convolutional, and continuous-time models with linear state space layers,” Advances in neural information processing systems, vol. 34, pp. 572–585, 2021

    2021

  78. [86]

    Long expressive memory for sequence modeling,

    T. K. Rusch, S. Mishra, N. B. Erichson, and M. W. Mahoney, “Long expressive memory for sequence modeling,” arXiv preprint arXiv:2110.04744, 2021

    2021 arXiv

  79. [87]

    Deep independently recurrent neural network (indrnn),

    S. Li, W. Li, C. Cook, and Y . Gao, “Deep independently recurrent neural network (indrnn),” arXiv preprint arXiv:1910.06251, 2019

    1910 arXiv

  80. [88]

    Coupled oscillatory recurrent neural network (cornn): An accurate and (gradient) stable architecture for learning long time dependencies,

    T. Konstantin Rusch and S. Mishra, “Coupled oscillatory recurrent neural network (cornn): An accurate and (gradient) stable architecture for learning long time dependencies,” arXiv e-prints, pp. arXiv–2010, 2020

    2010

  81. [89]

    Lip- schitz recurrent neural networks,

    N. B. Erichson, O. Azencot, A. Queiruga, and M. W. Mahoney, “Lip- schitz recurrent neural networks,” arXiv preprint arXiv:2006.12070 , 2020

    2006 arXiv

  82. [90]

    Gating revisited: Deep multi-layer rnns that can be trained,

    M. O. Turkoglu, S. D’Aronco, J. D. Wegner, and K. Schindler, “Gating revisited: Deep multi-layer rnns that can be trained,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 8, pp. 4081– 4092, 2021

    2021

  83. [91]

    Ckconv: Continuous kernel convolution for sequen- tial data,

    D. W. Romero, A. Kuzina, E. J. Bekkers, J. M. Tomczak, and M. Hoogendoorn, “Ckconv: Continuous kernel convolution for sequen- tial data,” arXiv preprint arXiv:2102.02611, 2021

    2021 arXiv

  84. [92]

    Recurrent batch normalization,

    T. Cooijmans, N. Ballas, C. Laurent, Ç. Gülçehre, and A. Courville, “Recurrent batch normalization,” arXiv preprint arXiv:1603.09025 , 2016

    2016 arXiv

  85. [93]

    Unitary evolution recurrent neural networks,

    M. Arjovsky, A. Shah, and Y . Bengio, “Unitary evolution recurrent neural networks,” in International Conference on Machine Learning , 2016, pp. 1120–1128

    2016

  86. [94]

    Long short-term memory,

    S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural computation, vol. 9, no. 8, pp. 1735–1780, 1997

    1997

  87. [95]

    MNIST handwritten digit database,

    Y . LeCun and C. Cortes, “MNIST handwritten digit database,” 2010

    2010

  88. [96]

    3d shapenets: A deep representation for volumetric shapes,

    Z. Wu, S. Song, A. Khosla, F. Yu, L. Zhang, X. Tang, and J. Xiao, “3d shapenets: A deep representation for volumetric shapes,” inProceedings of the IEEE conference on computer vision and pattern recognition , 2015, pp. 1912–1920

    2015

  89. [97]

    Shapenet: An information- rich 3d model repository,

    A. X. Chang, T. Funkhouser, L. Guibas, P. Hanrahan, Q. Huang, Z. Li, S. Savarese, M. Savva, S. Song, H. Suet al., “Shapenet: An information- rich 3d model repository,” arXiv preprint arXiv:1512.03012, 2015

    2015 arXiv

  90. [98]

    3dmatch: Learning local geometric descriptors from rgb-d reconstruc- tions,

    A. Zeng, S. Song, M. Nießner, M. Fisher, J. Xiao, and T. Funkhouser, “3dmatch: Learning local geometric descriptors from rgb-d reconstruc- tions,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1802–1811

    2017

  91. [99]

    Fully convolutional geometric fea- tures,

    C. Choy, J. Park, and V . Koltun, “Fully convolutional geometric fea- tures,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 8958–8966. IEEE TRANSACTION ON PATTERN ANAL YSIS AND MACHINE INTELLIGENCE 12

    2019

  92. [100]

    D3feat: Joint learning of dense detection and description of 3d local features,

    X. Bai, Z. Luo, L. Zhou, H. Fu, L. Quan, and C.-L. Tai, “D3feat: Joint learning of dense detection and description of 3d local features,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 6359–6367

    2020

  93. [101]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  94. [102]

    A method for registration of 3-d shapes,

    P. Besl and N. D. McKay, “A method for registration of 3-d shapes,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 14, no. 2, pp. 239–256, 1992

    1992

  95. [103]

    Go-icp: A globally optimal solution to 3d icp point-set registration,

    J. Yang, H. Li, D. Campbell, and Y . Jia, “Go-icp: A globally optimal solution to 3d icp point-set registration,” IEEE transactions on pattern analysis and machine intelligence, vol. 38, no. 11, pp. 2241–2254, 2015

    2015

  96. [104]

    Fast global registration,

    Q.-Y . Zhou, J. Park, and V . Koltun, “Fast global registration,” vol. 9906, 10 2016

    2016

  97. [105]

    Pointnetlk: Robust efficient point cloud registration using pointnet,

    Y . Aoki, H. Goforth, R. Arun Srivatsan, and S. Lucey, “Pointnetlk: Robust efficient point cloud registration using pointnet,” in The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019

    2019

  98. [106]

    Deepgmr: Learning latent gaussian mixture models for registration,

    W. Yuan, B. Eckart, K. Kim, V . Jampani, D. Fox, and J. Kautz, “Deepgmr: Learning latent gaussian mixture models for registration,” in European conference on computer vision. Springer, 2020, pp. 733– 750

    2020

  99. [107]

    Unsupervised point cloud registration via salient points analysis (spa),

    P. Kadam, M. Zhang, S. Liu, and C. C. J. Kuo, “Unsupervised point cloud registration via salient points analysis (spa),” 2020. [Online]. Available: https://arxiv.org/abs/2009.01293

    2020 arXiv

  100. [108]

    End-to-end 3d point cloud learning for registration task using virtual correspondences,

    H. Wei, Z. Qiao, Z. Liu, C. Suo, P. Yin, Y . Shen, H. Li, and H. Wang, “End-to-end 3d point cloud learning for registration task using virtual correspondences,” in 2020 IEEE/RSJ International Conference on In- telligent Robots and Systems (IROS). IEEE, 2020, pp. 2678–2683

    2020

  101. [109]

    Graphite: Graph-induced feature extraction for point cloud registration,

    M. Saleh, S. Dehghani, B. Busam, N. Navab, and F. Tombari, “Graphite: Graph-induced feature extraction for point cloud registration,” in 2020 International Conference on 3D Vision (3DV) . IEEE, 2020, pp. 241– 251

    2020

  102. [110]

    Corsnet: 3d point cloud registration by deep neural network,

    A. Kurobe, Y . Sekikawa, K. Ishikawa, and H. Saito, “Corsnet: 3d point cloud registration by deep neural network,” IEEE Robotics and Automation Letters, vol. 5, no. 3, pp. 3960–3966, 2020

    2020

  103. [111]

    Coherent point drift net- works: Unsupervised learning of non-rigid point set registration,

    L. Wang, X. Li, J. Chen, and Y . Fang, “Coherent point drift net- works: Unsupervised learning of non-rigid point set registration,” arXiv preprint arXiv:1906.03039, 2019

    1906 arXiv

  104. [112]

    Learning generalized non-rigid multimodal biomedical image registration from generic point set data,

    Z. M. Baum, T. Ungi, C. Schlenger, Y . Hu, and D. C. Barratt, “Learning generalized non-rigid multimodal biomedical image registration from generic point set data,” 2022. [Online]. Available: https://arxiv.org/abs/2207.10994

    2022 arXiv

  105. [113]

    Rar: Region-aware point cloud registration,

    Y . Hao and Y . Fang, “Rar: Region-aware point cloud registration,”

  106. [114]

    Feature interactive representation for point cloud registration,

    B. Wu, J. Ma, G. Chen, and P. An, “Feature interactive representation for point cloud registration,” in ICCV, 2021

    2021

  107. [115]

    Sdfreg: Learning signed distance functions for point cloud registration,

    L. Zhang, Y . Wang, Z. Lu, and L. Feng, “Sdfreg: Learning signed distance functions for point cloud registration,” arXiv preprint arXiv:2304.08929, 2023

    2023 arXiv

  108. [116]

    Self-supervised point cloud registration with deep versatile descriptors for intelligent driving,

    D. Liu, C. Chen, C. Xu, R. C. Qiu, and L. Chu, “Self-supervised point cloud registration with deep versatile descriptors for intelligent driving,” Transactions on Intelligent Transportation Systems, 2023

    2023

  109. [117]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in European conference on computer vision. Springer, 2014, pp. 740–755

    2014

  110. [118]

    MAGSAC: marginalizing sample consensus,

    D. Barath, J. Matas, and J. Noskova, “MAGSAC: marginalizing sample consensus,” inConference on Computer Vision and Pattern Recognition, 2019

    2019

  111. [119]

    Lf-net: Learning local features from images,

    Y . Ono, E. Trulls, P. Fua, and K. M. Yi, “Lf-net: Learning local features from images,” Advances in neural information processing systems , vol. 31, 2018

    2018

  112. [120]

    Localtrans: A multiscale local transformer network for cross-resolution homography estimation,

    R. Shao, G. Wu, Y . Zhou, Y . Fu, L. Fang, and Y . Liu, “Localtrans: A multiscale local transformer network for cross-resolution homography estimation,” in IEEE Conference on Computer Vision (ICCV 2021) , 2021. Ziming Zhang is an assistant professor at Worcester Polytechnic In...

    2021

  113. [2017]

    Zhang received his PhD in 2013 from Oxford Brookes University, UK, under the super- vision of Prof

    Dr. Zhang received his PhD in 2013 from Oxford Brookes University, UK, under the super- vision of Prof. Philip H. S. Torr. His research interests lie in computer vision and machine learning. He won the R&D 100 Award 2018. Yuping Shao received his M.S. in Electrical and Compute...

    2013

  114. [2021]

    Available: https://arxiv.org/abs/2110.03544

    [Online]. Available: https://arxiv.org/abs/2110.03544

    arXiv

  115. [2024]

    Haichong Zhang is an Associate Professor at Worcester Polytechnic Institute (WPI)

    His research interests include 3D point cloud and medical imaging applications. Haichong Zhang is an Associate Professor at Worcester Polytechnic Institute (WPI). He is the founding director of the Medical Fron- tier Ultrasound Imaging and Robotic Instru- mentation (FUSION) La...

Pith tools

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