Pith. sign in

REVIEW 3 major objections 5 minor 43 references

Developing Creative AI to Generate Sculptural Objects

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a union-and-resample step turns naive DeepDream-style point-cloud hallucination into dense, 3D-printable sculptures, and that segmenting the cloud lets different regions morph toward different targets.

desk verdict A clearly written artist-researcher paper with a genuinely new algorithmic trick (union-based repopulation during DeepDream-style point-cloud gradient ascent), but the headline claim of solving sparsity overreaches: the density gain is shown qualitatively, not measured, and the unspecified downsampling step is the load-bearing weak point. read the letter →

arxiv 1908.07587 v1 pith:OIMEZ4NA submitted 2019-08-20 cs.LG cs.AIcs.GRstat.ML

Songwei Ge , Austin Dill , Eunsu Kang , Chun-Liang Li , Lingyao Zhang , Manzil Zaheer , Barnabas Poczos This is my paper · ORCID
classification cs.LGcs.AIcs.GRstat.ML
keywords creativeAIpointcloudgenerationDeepDream3Dprintinggenerativeartsculpturegradientascentsegmentation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that DeepDream, the technique that makes neural networks hallucinate psychedelic images, can be moved to 3D point clouds to generate original sculptural objects that can actually be printed. Simply running gradient ascent on a point-cloud classifier produces sparse, hole-ridden clouds that cannot be meshed. The authors' Amalgamated DeepDream algorithm fixes this by periodically re-adding the original points to the deformed cloud and downsampling, using the fact that point clouds are permutation-invariant sets; this yields dense, printable objects while keeping the classifier-driven deformation. A second algorithm, Partitioned DeepDream, segments an object and runs the process per segment, so different parts of one object can morph toward different target classes. The authors physically print the results and exhibit them in an interactive installation.

What carries the argument

The load-bearing mechanism is the amalgamation operation in Algorithm 1: after every ten gradient-ascent steps, the current deformed point cloud is united with the original sub-cloud and then downsampled to a fixed size. This single operation solves both failure modes—local holes are filled by re-introduced original points, and global density is set by choosing how many subsets to concatenate. The second mechanism is the segmentation-plus-standardization loop of PDD: each segment is scaled to zero mean and unit variance, fed through ADD as though it were a whole object, then scaled back and re-unioned; this makes localized, multiple morphs possible. The guiding algorithmic assumption is that gradient ascent on a permutation-invariant classifier's class activation is a useful deformation primitive.

What would settle it

Run the same ADD pipeline with a classifier whose weights are randomly initialized (or trained on permuted labels) while keeping all other settings identical; if the resulting point clouds are still judged sculptural and print as well, then the deformation does not depend on learned class structure, and the 'creative' claim loses its stated mechanism.

Watch

Extended reading notes

Core claim

The authors' central claim is that the failure mode of naive point-cloud DeepDream—local holes and an overall too-thin point set—is not an inherent limit of gradient ascent but a resolvable engineering problem. ADD (Amalgamated DeepDream) runs gradient ascent on a trained point-cloud classifier for a fixed target class, and every ten steps takes the union of the current deformed cloud with the original input cloud, then downsamples back to the model's capacity. Because point clouds are sets, this union is a single, mathematically clean operation; it continually reintroduces original points so that density is preserved while the gradient updates still reshape the object. After the transformation, all per-subset results are themselves unioned, yielding a cloud as dense as the input allows. PDD (Partitioned DeepDream) applies ADD separately to standardized segments of an input—obtained either from k-means or from manual partitions—and then re-unions the transformed segments, so one object can move toward several target classes at once. The paper carries the pipeline through to physical realization: Poisson-disk sub-sampling, ball-pivoting surface reconstruction, solidification, and 3D printing.

Load-bearing premise

That maximizing a trained classifier's class activation on a point cloud produces aesthetically meaningful, coherent forms rather than just adversarial noise.

Editorial extensions

If this is right

  • ADD-generated point clouds are dense enough for meshing and 3D printing, so DeepDream-style hallucination is no longer confined to 2D images.
  • PDD gives artists controllable local morphing: one region of an object can be deformed to one class while another region is preserved or morphed to a different class.
  • The union trick sets point-cloud density independently of the classifier's input capacity, so the output's point count is bounded only by the input cloud.
  • The printed sculptures and installation demonstrate that machine-learning generation can enter the physical art world, not just the screen.
  • The authors show these outputs are not simple copies of training categories, retaining input identity while acquiring target features.

Reading between the lines

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

  • The same periodic-union-and-downsample trick could be transplanted to any gradient-based point-cloud edit (adversarial perturbation, style transfer, part-aware editing) wherever density loss is a problem.
  • PDD with spatial block segmentation hints at an interactive 'paint-a-target-class' interface for sculptors, but the paper does not implement user-directed region selection beyond the tested partitions.
  • A quantitative test of 'creativity' is left open: one could measure shape diversity (e.g., distance from input and from training-set categories) across ADD outputs and compare with GAN or autoencoder baselines.
  • Because the method only needs a differentiable classifier, it should carry over to voxel- or mesh-based classifiers, though the amalgamation operation would need a different definition there.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes two algorithms for generating 3D point clouds for sculptural art: Amalgamated DeepDream (ADD) and Partitioned DeepDream (PDD). ADD iteratively applies gradient ascent on a fixed DeepSet classifier's class activations, periodically taking the union of the evolving point cloud with the original input and down-sampling to control point count, with the goal of avoiding the sparsity that plagues a naive DeepDream extension. PDD applies ADD to individually standardized segments obtained by k-means or manual segmentation, allowing different transformations on different object parts. The authors claim that ADD 'solves the sparsity problem' and that the outputs are creative and printable; they demonstrate physical 3D-printed sculptures and describe their integration into an interactive art installation.

Significance. If the density claim is substantiated, this is a valuable application of classifier-gradient methods to 3D point-cloud generation, and it addresses an underexplored intersection of machine learning and sculpture. The union-and-resample idea is simple and plausible, the algorithms are stated in pseudocode, the classifier is fixed and pretrained, and the reported physical sculptures provide concrete evidence that the full reconstruction pipeline can work. The work also has clear artistic relevance. The main weakness is that the central technical claim—that ADD solves the sparsity problem and yields generally printable objects—is supported almost entirely by qualitative figures and a single pipeline instance, with no quantitative density or reconstruction evaluation.

major comments (3)
  1. [Algorithm 1 (ADD)] Algorithm 1 specifies 'down-sample xt to fix the number of points' without identifying the downsampling method. If this is uniform random resampling, then the union of xt with x followed by random subsampling back to the original size does not guarantee preservation of local density: transformed points can migrate away from initially covered regions, and random subsampling may thin those regions further. Because the paper's central claim is that ADD solves the sparsity problem, the downsampler must be specified (e.g., farthest-point or Poisson-disk sampling) and its density-preservation property must be demonstrated. As written, the algorithm is not reproducible and the claim is not supported.
  2. [Realization of Sculptural Object, 'From points clouds to mesh'] The manuscript states in this section that 'ADD results in point clouds with a greater variance in sparsity than typical point clouds generated from real-world objects.' This is in tension with the abstract's claim that ADD 'solves the sparsity problem caused by the naive DeepDream-inspired approach.' The claim should be qualified and quantified: report local density statistics (e.g., k-nearest-neighbor distances or occupancy in a fixed grid) for naive DeepDream, ADD, and real-world point clouds, and connect these to reconstruction success. Without such a comparison, 'solves' is an overstatement.
  3. [Amalgamated DeepDream (ADD), 'Experiments and results'] The ADD experiments are purely visual. There is no quantitative comparison against the naive DeepDream baseline on the same input, no repeated runs to account for the random division of the input into subsets, and no measure of reconstruction quality such as the success rate of Ball-Pivoting meshing, mesh manifoldness, or number of connected components. Since printability is a central claim, a measurable criterion applied across several object-target pairs and seeds is needed; a single successful pipeline instance does not establish general printability.
minor comments (5)
  1. [Algorithm 1] The notation 'for x = X0,...,Xn do' is nonstandard; it should be written as 'for x in {X0,...,Xn} do' or similar, and the scoping of xT should be clarified.
  2. [Amalgamated DeepDream (ADD)] The statement that 'we can generate as dense objects as we want as long as the input object allows' is informal; the authors should define density and state the relationship between input point count, subset size, and final density.
  3. [Figure 8 and ADD experiments] The naive DeepDream result in Figure 8 lacks the parameter settings (number of iterations, learning rate, point count) used to produce the sparse result, which is needed for a fair comparison with ADD.
  4. [Introduction] There is a typo in the introduction: 'V ox2Net' should be 'Vox2Net'.
  5. [Conclusion] The conclusion acknowledges the lack of an objective metric for creativity; this limitation could be stated earlier and the qualitative claims matched to it.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: ADD/PDD are algorithmic constructions whose outputs are produced by gradient ascent on a fixed classifier, not by fitting or self-citation.

full rationale

The central claim is that ADD solves the naive DeepDream sparsity problem and yields printable point clouds. The algorithm is specified as repeated gradient updates xt = xt-1 + gamma * grad_x f_theta(xt-1; a), with periodic union with the original subset and down-sampling, followed by amalgamating transformed subsets. The output is generated, not fitted: no parameter is tuned to match the resulting sculptures, and the DeepSet classifier is fixed from prior published work. The self-citations (Deep Sets [41], Point Cloud GAN [23]) provide the base architecture and background interpolation examples; they are not invoked as an authority to prove that ADD solves sparsity or that the outputs are printable. No uniqueness theorem, ansatz, or fitted 'prediction' is imported. The paper's acknowledgment that 'ADD results in point clouds with a greater variance in sparsity than typical point clouds generated from real-world objects' weakens the strength of the sparsity claim, but that is an evidence or over-claim issue, not a circularity. The method is self-contained in the sense that its outputs are consequences of the stated update rule applied to a trained network; whether those outputs are aesthetically creative or reliably printable is an empirical question, not a definitional equivalence.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

All free parameters are hand-chosen hyperparameters of the algorithm, not quantities fitted to a target result. The paper introduces no new physical or conceptual entities. The key domain assumptions are that classifier gradients produce meaningful aesthetics and that simple geometric operations (union, down-sampling, Ball-Pivoting reconstruction) preserve enough shape information for printing. These assumptions are plausible but only qualitatively validated.

free parameters (6)
  • Learning rate gamma = 1
    Set for all ADD and PDD experiments; controls the step size of each gradient update.
  • Number of iterations T = 100
    Applied to each subset in ADD; determines the extent of deformation.
  • Down-sample interval = 10 iterations
    ADD re-unions each transformed subset with the original input every 10 iterations, then down-samples to a fixed point count.
  • Sampled points per object = 10000
    Randomly sampled from ModelNet40/ShapeNet CAD surfaces as input and for meshing.
  • Number of k-means segments = 4 for airplane example; unspecified generally
    PDD requires the number of segments as an input; Figure 13 shows an airplane split into 4 clusters.
  • Block partition grid = 3x3x2
    Manual block segmentation divides the bounding box into 18 blocks for PDD.
assumptions (4)
  • domain assumption DeepSet can serve as a trained classifier fθ whose class activations provide useful gradient signals for 3D shape deformation.
    Invoked in the 'Inspiration: DeepDream for Point Clouds' section and Algorithms 1-2; the entire method's output quality depends on this.
  • domain assumption Uniformly sampling 10000 points from CAD model surfaces adequately represents object geometry for the deep network and for mesh reconstruction.
    Data preprocessing paragraph states surfaces are uniformly sampled; all results depend on this representation.
  • domain assumption Ball-Pivoting after Poisson Disk Sub-sampling yields a faithful mesh of the generated point cloud for 3D printing.
    Section 'From points clouds to mesh' states this choice was made after visual trials; no quantitative reconstruction error given.
  • ad hoc to paper Set union of point clouds followed by down-sampling preserves the object's relevant shape features while filling holes.
    Amalgamation in Algorithm 1 is the paper's proposed mechanism; its effectiveness is shown only visually.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Developing Creative AI to Generate Sculptural Objects." pith.science (2026). https://pith.science/paper/OIMEZ4NA

@misc{pith2026190807587,
  author       = {Pith},
  title        = {Pith review of: Developing Creative AI to Generate Sculptural Objects},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OIMEZ4NA}},
  note         = {Machine review of arXiv:1908.07587}
}
read the original abstract

We explore the intersection of human and machine creativity by generating sculptural objects through machine learning. This research raises questions about both the technical details of automatic art generation and the interaction between AI and people, as both artists and the audience of art. We introduce two algorithms for generating 3D point clouds and then discuss their actualization as sculpture and incorporation into a holistic art installation. Specifically, the Amalgamated DeepDream (ADD) algorithm solves the sparsity problem caused by the naive DeepDream-inspired approach and generates creative and printable point clouds. The Partitioned DeepDream (PDD) algorithm further allows us to explore more diverse 3D object creation by combining point cloud clustering algorithms and ADD.

Figures

Figures reproduced from arXiv: 1908.07587 by the authors.

Figure 1
Figure 1. Sculpture generated by creative AI, PDD. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 5
Figure 5. Encoding and decoding in latent space final object Xˆ. Specifically, an autoencoder is a kind of feedforward neural network used for dimensionality reduc￾tion whose hidden units can be viewed as latent codes which capture the most important aspects of the object [30]. A rep￾resentation of the process of encoding to and decoding from latent space can be seen in [PITH_FULL_IMAGE:figures/full_fig_p002_5.png] view at source ↗
Figure 2
Figure 2. ModelNet40 [38] [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: An example of generative model. Here a low-dimensional latent space h of the original ob￾ject space x is learned in accordance with the underlying probability distribution. p(x) = Z h p(h)p(x|h)dh (1) A discriminator is usually used to help distinguish the gener￾ated o…
Figure 7
Figure 7. Figure 7: 2D DeepDream visualized some features that are captured by fθ(x; a). Algorithmically, DeepDream iteratively modifies x via a gradient update with a certain learning rate γ. xt = xt−1 + γ∇xfθ(x; a). (2) [PITH_FULL_IMAGE:figures/full_fig_p003_7.png]
Figure 8
Figure 8. Figure 8: Naive DeepDream with undesirable sparse surface. An extension of Deep￾Dream is to use a clas￾sification network as fθ and we replace a with the outputs of the final layer corresponding to certain labels. This is related to adversarial attack [33] and unrestricted adver…
Figure 10
Figure 10. Figure 10: Transforming a bottle into a cone via ADD. [PITH_FULL_IMAGE:figures/full_fig_p003_10.png]
Figure 13
Figure 13. Figure 13: Segmentation results of an airplane with 4 clusters. [PITH_FULL_IMAGE:figures/full_fig_p004_13.png]
Figure 12
Figure 12. Figure 12: ADD with dual and triple object input [PITH_FULL_IMAGE:figures/full_fig_p004_12.png]
Figure 15
Figure 15. Figure 15: PDD with guitar body only Instead, we explore machine learning methods which can au￾tomatically divide the object into several meaningful parts. For example, the segmentation results of an airplane with k￾means are displayed in [PITH_FULL_IMAGE:figures/full_fig_p005_…
Figure 17
Figure 17. Figure 17: PDD with block segmented objects targeting random categories [PITH_FULL_IMAGE:figures/full_fig_p005_17.png]
Figure 16
Figure 16. Figure 16: PDD on uniformly segmented objects out remains that of a guitar. PDD with manual segmentation strategies In addition to automatic segmentation algorithms, two manual segmenta￾tion strategies have been tested. First, for highly symmet￾ric objects it is reasonable to un…
Figure 21
Figure 21. Figure 21: Created sculpture from ADD [PITH_FULL_IMAGE:figures/full_fig_p006_21.png]
Figure 22
Figure 22. Figure 22: Sketch of Aural Fauna: Illuminato art installation [PITH_FULL_IMAGE:figures/full_fig_p006_22.png]
Figure 20
Figure 20. Figure 20: Screen capture from Cura-Lulzbot software for 3d [PITH_FULL_IMAGE:figures/full_fig_p006_20.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 34 canonical work pages

  1. [1]

    Achlioptas, P.; Diamanti, O.; Mitliagkas, I.; and Guibas, L. 2017. Learning representations and generative models for 3d point clouds. arXiv preprint arXiv:1707.02392

    arXiv 2017

  2. [2]

    Ballard, D. H. 1981. Generalizing the hough transform to detect arbitrary shapes. Pattern recognition 13(2):111– 122

    work page 1981

  3. [3]

    Bernardini, F.; Mittleman, J.; Rushmeier, H.; Silva, C.; and Taubin, G. 1999. The ball-pivoting algorithm for sur- face reconstruction. IEEE transactions on visualization and computer graphics 5(4):349–359

    work page 1999

  4. [4]

    J., and Jain, R

    Besl, P. J., and Jain, R. C. 1988. Segmentation through variable-order surface fitting. IEEE Transactions on Pat- tern Analysis and Machine Intelligence 10(2):167–192

    work page 1988

  5. [5]

    Bidgoli, A., and Veloso, P. 2018. Deepcloud: The appli- cation of a data-driven, generative model in design

    work page 2018

  6. [6]

    B.; Carlini, N.; Zhang, C.; Olsson, C.; Chris- tiano, P.; and Goodfellow, I

    Brown, T. B.; Carlini, N.; Zhang, C.; Olsson, C.; Chris- tiano, P.; and Goodfellow, I. 2018. Unrestricted adversarial examples. arXiv preprint arXiv:1809.08352

    arXiv 2018

  7. [7]

    Carter, S., and Nielsen, M. 2017. Using artificial intelli- gence to augment human intelligence. Distill

    work page 2017

  8. [8]

    Cazals, F., and Giesen, J. 2006. Delaunay triangulation based surface reconstruction. In Effective computational geometry for curves and surfaces . Springer. 231–276

    work page 2006

Show all 43 references
  1. [9]

    X.; Funkhouser, T.; Guibas, L.; Hanrahan, P.; Huang, Q.; Li, Z.; Savarese, S.; Savva, M.; Song, S.; Su, H.; et al

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

    2015 arXiv

  2. [10]

    Cignoni, P.; Callieri, M.; Corsini, M.; Dellepiane, M.; Ganovelli, F.; and Ranzuglia, G. 2008. MeshLab: an Open-Source Mesh Processing Tool. In Scarano, V .; Chiara, R. D.; and Erra, U., eds., Eurographics Italian Chapter Conference. The Eurographics Association

    2008

  3. [11]

    http://egorkraft.art/#cas

    Content aware studies. http://egorkraft.art/#cas

  4. [12]

    Corsini, M.; Cignoni, P.; and Scopigno, R. 2012. Efficient and flexible sampling with blue noise prop- erties of triangular meshes. IEEE Transaction on Visualization and Computer Graphics 18(6):914–924. http://doi.ieeecomputersociety.org/10.1109/TVCG.2012.34

    2012 doi

  5. [13]

    Elgammal, A.; Liu, B.; Elhoseiny, M.; and Mazzone, M

  6. [14]

    A., and Bolles, R

    Fischler, M. A., and Bolles, R. C. 1981. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Communi- cations of the ACM 24(6):381–395

    1981

  7. [15]

    Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; and Bengio, Y

  8. [16]

    Grilli, E.; Menna, F.; and Remondino, F. 2017. A re- view of point clouds segmentation and classification al- gorithms. The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences 42:339

    2017

  9. [17]

    G.; Russell, B

    Groueix, T.; Fisher, M.; Kim, V . G.; Russell, B. C.; and Aubry, M. 2018. Atlasnet: A papier-mache ap- proach to learning 3d surface generation. arXiv preprint arXiv:1802.05384

    2018 arXiv

  10. [18]

    Guo, B.; Menon, J.; and Willette, B. 1997. Surface reconstruction using alpha shapes. In Computer Graphics F orum, volume 16, 177–190. Wiley Online Library

    1997

  11. [19]

    Hoppe, H.; DeRose, T.; Duchamp, T.; McDonald, J.; and Stuetzle, W. 1992. Surface reconstruction from unor- ganized points

    1992

  12. [20]

    Kato, H.; Ushiku, Y .; and Harada, T. 2018. Neural 3d mesh renderer. In CVPR

    2018

  13. [21]

    Kazhdan, M., and Hoppe, H. 2013. Screened poisson surface reconstruction. ACM Transactions on Graphics (TOG) 32(3):29

    2013

  14. [22]

    Lehman, J.; Risi, S.; and Clune, J. 2016. Creative gen- eration of 3d objects with deep learning and innovation en- gines. In Proceedings of the 7th International Conference on Computational Creativity

    2016

  15. [23]

    Li, C.-L.; Zaheer, M.; Zhang, Y .; Poczos, B.; and Salakhutdinov, R. 2018. Point cloud gan. arXiv preprint arXiv:1810.05795

    2018 arXiv

  16. [24]

    D.; Tao, M.; Li, C.-L.; Nowrouzezahrai, D.; and Jacobson, A

    Liu, H.-T. D.; Tao, M.; Li, C.-L.; Nowrouzezahrai, D.; and Jacobson, A. 2018. Adversarial geometry and lighting using a differentiable renderer. arXiv preprint arXiv:1808.02651

    2018 arXiv

  17. [25]

    Lu, X.; Yao, J.; Tu, J.; Li, K.; Li, L.; and Liu, Y . 2016. Pairwise linkage for point cloud segmentation. ISPRS An- nals of Photogrammetry, Remote Sensing & Spatial Infor- mation Sciences 3(3)

    2016

  18. [26]

    Nguyen, A., and Le, B. 2013. 3d point cloud segmenta- tion: A survey. In RAM, 225–230

    2013

  19. [27]

    Nima, D.; Luca, S.; and Mauro, M. 2018. V ox2net: From 3d shapes to network sculptures. In NIPS

    2018

  20. [28]

    O., and Tyka, M

    Posted by Alexander Mordvintsev, C. O., and Tyka, M. Inceptionism: Going deeper into neural net- works. https://ai.googleblog.com/2015/06/inceptionism- going-deeper-into-neural.html

    2015

  21. [29]

    R.; Su, H.; Mo, K.; and Guibas, L

    Qi, C. R.; Su, H.; Mo, K.; and Guibas, L. J. 2017. Pointnet: Deep learning on point sets for 3d classification and segmentation. CVPR

    2017

  22. [30]

    Robert, C. 2014. Machine learning, a probabilistic per- spective

    2014

  23. [31]

    D., and Devy, M

    Sappa, A. D., and Devy, M. 2001. Fast range image segmentation by an edge detection strategy. In3-D Digital Imaging and Modeling, 2001. Proceedings. Third Interna- tional Conference on, 292–299. IEEE

    2001

  24. [32]

    Schmidt, R., and Singh, K. 2010. Meshmixer: an in- terface for rapid mesh composition. In ACM SIGGRAPH 2010 Talks, 6. ACM

    2010

  25. [33]

    Szegedy, C.; Zaremba, W.; Sutskever, I.; Bruna, J.; Erhan, D.; Goodfellow, I.; and Fergus, R. 2013. In- triguing properties of neural networks. arXiv preprint arXiv:1312.6199

    2013 arXiv

  26. [34]

    J.; Gedikli, S.; Rusu, R

    Trevor, A. J.; Gedikli, S.; Rusu, R. B.; and Christensen, H. I. 2013. Efficient organized point cloud segmentation with connected components. Semantic Perception Map- ping and Exploration (SPME)

    2013

  27. [35]

    W.; and Kavukcuoglu, K

    Van Den Oord, A.; Dieleman, S.; Zen, H.; Simonyan, K.; Vinyals, O.; Graves, A.; Kalchbrenner, N.; Senior, A. W.; and Kavukcuoglu, K. 2016. Wavenet: A gener- ative model for raw audio. In SSW

    2016

  28. [36]

    Vieira, M., and Shimada, K. 2005. Surface mesh seg- mentation and smooth surface extraction through region growing. Computer aided geometric design 22(8):771– 792

    2005

  29. [37]

    A., and Arabnia, H

    Wani, M. A., and Arabnia, H. R. 2003. Parallel edge- region-based segmentation algorithm targeted at reconfig- urable multiring network. The Journal of Supercomputing 25(1):43–62

    2003

  30. [38]

    Wu, Z.; Song, S.; Khosla, A.; Yu, F.; Zhang, L.; Tang, X.; and Xiao, J. 2015. 3d shapenets: A deep representation for volumetric shapes. In CVPR

    2015

  31. [39]

    Wu, J.; Zhang, C.; Xue, T.; Freeman, B.; and Tenen- baum, J. 2016. Learning a probabilistic latent space of object shapes via 3d generative-adversarial modeling. In NIPS

    2016

  32. [40]

    Yang, Y .; Feng, C.; Shen, Y .; and Tian, D. 2018. Fold- ingnet: Point cloud auto-encoder via deep grid deforma- tion. In CVPR, volume 3

    2018

  33. [41]

    R.; and Smola, A

    Zaheer, M.; Kottur, S.; Ravanbakhsh, S.; Poczos, B.; Salakhutdinov, R. R.; and Smola, A. J. 2017. Deep sets. In NIPS. Authors Biographies Songwei Ge is a Masters student in the Computational Biol- ogy Department at Carnegie Mellon University. Austin Dill is a Masters student i...

    2017

  34. [2014]

    Generative adversarial nets. In NIPS

  35. [2017]

    arXiv preprint arXiv:1706.07068

    Can: Creative adversarial networks, generating” art” by learning about styles and deviating from style norms. arXiv preprint arXiv:1706.07068

    arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.