CCN: Publications

Filter by Group

Computational Vision

Neural Circuits and Algorithms

NeuroAI and Geometric Data Analysis

Statistical Analysis of Neural Data

181 Publications

Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices

C. Chun, S. Chung , D. D. Lee

arXiv:2410.17998

Export Citation

October 23, 2024

Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices

@article{Chun_Chung_arxiv:2410.17998_2024,
title = {{Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices}},
author = {Chanwoo Chun and SueYeon Chung and Daniel D. Lee},
abstract = {Analyzing the structure of sampled features from an input data distribution is challenging when constrained by limited measurements in both the number of inputs and features. Traditional approaches often rely on the eigenvalue spectrum of the sample covariance matrix derived from finite measurement matrices; however, these spectra are sensitive to the size of the measurement matrix, leading to biased insights. In this paper, we introduce a novel algorithm that provides unbiased estimates of the spectral moments of the kernel integral operator in the limit of infinite inputs and features from finitely sampled measurement matrices. Our method, based on dynamic programming, is efficient and capable of estimating the moments of the operator spectrum. We demonstrate the accuracy of our estimator on radial basis function (RBF) kernels, highlighting its consistency with the theoretical spectra. Furthermore, we showcase the practical utility and robustness of our method in understanding the geometry of learned representations in neural networks.},
acrosscenters = {no},
external = {yes},
highlight = {no},
eprint = {2410.17998},
archivePrefix = {arxiv},
year_posted = {2024},
center = {ccn},
OpenAccessURL = {https://arxiv.org/abs/2410.17998},
Journal = {arXiv e-prints},
year = {2024}
}

Download Citation (BibTex)

High-fidelity Image Restoration of Large 3D Electron Microscopy Volume

Yuri Kreinin, P. Gunn, D. Chklovskii, J. Wu

bioRxiv:2023.09.14.557785

Microscopy and Microanalysis, 30: 889-902

Export Citation

October 18, 2024

High-fidelity Image Restoration of Large 3D Electron Microscopy Volume

"@article{Gunn_Wu_biorxiv:2023.09.14.557785_MIMI_2024,
title = {{High-fidelity Image Restoration of Large 3D Electron Microscopy Volume}},
author = {Kreinin, Yuri and Gunn, Pat and Chklovskii, Dmitri and Wu, Jingpeng},
abstract = {Volume electron microscopy (VEM) is an essential tool for studying biological structures. Due to the challenges of sample preparation and continuous volumetric imaging, image artifacts are almost inevitable. Such image artifacts complicate further processing both for automated computer vision methods and human experts. Unfortunately, the widely used contrast limited adaptive histogram equalization (CLAHE) can alter the essential relative contrast information about some biological structures. We developed an image-processing pipeline to remove the artifacts and enhance the images without CLAHE. We apply our method to VEM datasets of a Microwasp head. We demonstrate that our method restores the images with high fidelity while preserving the original relative contrast. This pipeline is adaptable to other VEM datasets.},
acrosscenters = {no},
external = {yes},
highlight = {no},
eprint = {https://www.biorxiv.org/content/10.1101/2023.09.14.557785v1},
archivePrefix = {BioRxiv},
year_posted = {2023},
center = {CCN},
OpenAccessURL = {https://www.biorxiv.org/content/early/2023/09/17/2023.09.14.557785},
year = {2024},
journal = MIMI,
volume = {30},
number = {5},
pages = {889-902},
month = {10}
}"

Download Citation (BibTex)

Neuronal and Behavioral Responses to Naturalistic Texture Images in Macaque Monkeys

Corey M. Ziemba, Robbe L.T. Goris, Gabriel M. Stine, Richard K. Perez, E. P. Simoncelli, J. A. Movshon

bioXriv:2024.02.22.581645

The Journal of Neuroscience, 44: e0349242024

Export Citation

October 16, 2024

Neuronal and Behavioral Responses to Naturalistic Texture Images in Macaque Monkeys

@article{Simoncelli_JON_2024,
title = {{Neuronal and Behavioral Responses to Naturalistic Texture Images in Macaque Monkeys}},
author = {Ziemba,  Corey M. and Goris,  Robbe L. T. and Stine,  Gabriel M. and Perez,  Richard K. and Simoncelli,  Eero P. and Movshon,  J. Anthony},
abstract = {The visual world is richly adorned with texture, which can serve to delineate important elements of natural scenes. In anesthetized macaque monkeys, selectivity for the statistical features of natural texture is weak in V1, but substantial in V2, suggesting that neuronal activity in V2 might directly support texture perception. To test this, we investigated the relation between single cell activity in macaque V1 and V2 and simultaneously measured behavioral judgments of texture. We generated stimuli along a continuum between naturalistic texture and phase-randomized noise and trained two macaque monkeys to judge whether a sample texture more closely resembled one or the other extreme. Analysis of responses revealed that individual V1 and V2 neurons carried much less information about texture naturalness than behavioral reports. However, the sensitivity of V2 neurons, especially those preferring naturalistic textures, was significantly closer to that of behavior compared with V1. The firing of both V1 and V2 neurons predicted perceptual choices in response to repeated presentations of the same ambiguous stimulus in one monkey, despite low individual neural sensitivity. However, neither population predicted choice in the second monkey. We conclude that neural responses supporting texture perception likely continue to develop downstream of V2. Further, combined with neural data recorded while the same two monkeys performed an orientation discrimination task, our results demonstrate that choice-correlated neural activity in early sensory cortex is unstable across observers and tasks, untethered from neuronal sensitivity, and therefore unlikely to directly reflect the formation of perceptual decisions.},
acrosscenters = {no},
external = {yes},
highlight = {no},
center = {ccn},
OpenAccessURL = {https://www.jneurosci.org/content/44/42/e0349242024},
Journal = JON,
publisher = {Society for Neuroscience},
volume = {44},
ISSN = {1529-2401},
DOI = {10.1523/jneurosci.0349-24.2024},
number = {42},
publisher = {Society for Neuroscience},
month = aug,
pages = {e0349242024},
year = {2024}
}

Download Citation (BibTex)

Feature-guided score diffusion for sampling conditional densities

Z. Kadkhodaie, S. Mallat, E. P. Simoncelli

arXiv:2410.11646

Export Citation

October 15, 2024

Feature-guided score diffusion for sampling conditional densities

@article{Kadkhodaie_Simoncelli_arxiv:2410.11646,
 title={{Feature-guided score diffusion for sampling conditional densities}},
 author={Kadkhodaie, Zahra and Mallat, St\'{{e}}phane and Simoncelli, Eero p},
 year_posted={2024},
 Journal = {arXiv e-prints},
 center = {CCN and CCM},
 acrosscenters = {yes},
 external = {no},
 highlight = {no},
year = {2024},
 abstract = {Score diffusion methods can learn probability densities from samples. The score of the noise-corrupted density is estimated using a deep neural network, which is then used to iteratively transport a Gaussian white noise density to a target density. Variants for conditional densities have been developed, but correct estimation of the corresponding scores is difficult. We avoid these difficulties by introducing an algorithm that guides the diffusion with a projected score. The projection pushes the image feature vector towards the feature vector centroid of the target class. The projected score and the feature vectors are learned by the same network. Specifically, the image feature vector is defined as the spatial averages of the channels activations in select layers of the network. Optimizing the projected score for denoising loss encourages image feature vectors of each class to cluster around their centroids. It also leads to the separations of the centroids. We show that these centroids provide a low-dimensional Euclidean embedding of the class conditional densities. We demonstrate that the algorithm can generate high quality and diverse samples from the conditioning class. Conditional generation can be performed using feature vectors interpolated between those of the training set, demonstrating out-of-distribution generalization.},
eprint={2410.11646},
archivePrefix={arXiv},
OpenAccessURL={https://arxiv.org/abs/2410.11646}
}

Download Citation (BibTex)

Comparing the local information geometry across several image representations

D. Lipshutz, J. Feather, S. Harvey, A. Williams, E. P. Simoncelli

UniReps: 2nd Edition of the Workshop on Unifying Representations in Neural Models

Export Citation

October 10, 2024

Comparing the local information geometry across several image representations

@INPROCEEDINGS{Lipshutz24d,
    AUTHOR=     "D Lipshutz* and J Feather* and S E Harvey and A H Williams and  E P Simoncelli",
  TITLE=        "Comparing the local information geometry across several image representations",
  BOOKTITLE=    "Proc. UniReps: Unifying Representations in Neural Models,
                 A workshop at Neural Information Processing Systems (NeurIPS)",
  ADDRESS=      "Vancouver",
  MONTH=        "Dec",
  YEAR=         2024,
  URL=          "https://openreview.net/forum?id=qwVRKFj8dM",
}

Download Citation (BibTex)

Equivalence between representational similarity analysis, centered kernel alignment, and canonical correlations analysis

A. Williams

bioXriv:2024.10.23.619871

UniReps: 2nd Edition of the Workshop on Unifying Representations in Neural Models, 2024

Export Citation

October 10, 2024

Equivalence between representational similarity analysis, centered kernel alignment, and canonical correlations analysis

"@article{Williams_biorxiv:2024.10.23.619871_UR2_2024,
title = {{Equivalence between representational similarity analysis, centered kernel alignment, and canonical correlations analysis}},
author = {Williams, Alex H.},
abstract = {Centered kernel alignment (CKA) and representational similarity analysis (RSA) of dissimilarity matrices are two popular methods for comparing neural systems in terms of representational geometry. Although they follow a conceptually similar approach, typical implementations of CKA and RSA tend to result in numerically different outcomes. Here, I show that these two approaches are largely equivalent once one incorporates a mean-centering step into RSA. This equivalence holds for both linear and nonlinear variants of these methods. These connections are simple to derive, but appear to have been thus far overlooked in the context of comparing neural representations. By unifying these measures, this paper hopes to simplify a complex and fragmented literature on this subject.Competing Interest StatementThe authors have declared no competing interest.},
acrosscenters = {no},
external = {no},
highlight = {no},
eprint = {https://www.biorxiv.org/content/early/2024/10/24/2024.10.23.619871.full.pdf},
archivePrefix = {bioRxiv},
year_posted = {2024},
Journal = {UR2},
center = {CCN},
OpenAccessURL = {https://www.biorxiv.org/content/early/2024/10/24/2024.10.23.619871},
year = {2024},
doi = {10.1101/2024.10.23.619871},
publisher = {Cold Spring Harbor Laboratory}
}"

Download Citation (BibTex)

What Representational Similarity Measures Imply about Decodable Information

S. Harvey, D. Lipshutz, A. Williams

arXiv:2411.08197

UniReps: 2nd Edition of the Workshop on Unifying Representations in Neural Models

Export Citation

October 10, 2024

What Representational Similarity Measures Imply about Decodable Information

@article{Harvey_Williams_arXiv:2411.08197_UR2_2024,
title = {{What Representational Similarity Measures Imply about Decodable Information}},
author = {Sarah E. Harvey and David Lipshutz and Alex H. Williams},
year_posted = {2024},
Journal = UR2,
center = {CCN},
acrosscenters = {no},
external = {no},
highlight = {no},
year = {2024},
abstract = {Neural responses encode information that is useful for a variety of downstream tasks. A common approach to understand these systems is to build regression models or ``decoders'' that reconstruct features of the stimulus from neural responses. Popular neural network similarity measures like centered kernel alignment (CKA), canonical correlation analysis (CCA), and Procrustes shape distance, do not explicitly leverage this perspective and instead highlight geometric invariances to orthogonal or affine transformations when comparing representations. Here, we show that many of these measures can, in fact, be equivalently motivated from a decoding perspective. Specifically, measures like CKA and CCA quantify the average alignment between optimal linear readouts across a distribution of decoding tasks. We also show that the Procrustes shape distance upper bounds the distance between optimal linear readouts and that the converse holds for representations with low participation ratio. Overall, our work demonstrates a tight link between the geometry of neural representations and the ability to linearly decode information. This perspective suggests new ways of measuring similarity between neural systems and also provides novel, unifying interpretations of existing measures.},
archivePrefix = {arXiv},
eprint = {2411.08197},
OpenAccessURL = {https://arxiv.org/abs/2411.08197}
}

Download Citation (BibTex)

Classification-Denoising Networks

Louis Thiry, F. Guth

arXiv:2410.03505

Export Citation

October 4, 2024

Classification-Denoising Networks

@article{Thiry_Guth_arXiv:2410.03505,
      title={{Classification-Denoising Networks}},
      author={Louis Thiry and Florentin Guth},
      year={2024},
    publisher={Cornell Tech},
    journal = {arXiv e-prints},
    doi={10.48550/arXiv.2410.03505},
    abstract={Image classification and denoising suffer from complementary issues of lack of robustness or partially ignoring conditioning information. We argue that they can be alleviated by unifying both tasks through a model of the joint probability of (noisy) images and class labels. Classification is performed with a forward pass followed by conditioning. Using the Tweedie-Miyasawa formula, we evaluate the denoising function with the score, which can be computed by marginalization and back-propagation. The training objective is then a combination of cross-entropy loss and denoising score matching loss integrated over noise levels. Numerical experiments on CIFAR-10 and ImageNet show competitive classification and denoising performance compared to reference deep convolutional classifiers/denoisers, and significantly improves efficiency compared to previous joint approaches. Our model shows an increased robustness to adversarial perturbations compared to a standard discriminative classifier, and allows for a novel interpretation of adversarial gradients as a difference of denoisers.},
      eprint={2410.03505},
      archivePrefix={arXiv},
      OpenAccessUrl={https://arxiv.org/abs/2410.03505}
}

Download Citation (BibTex)

A Rainbow in Deep Network Black Boxes

F. Guth, Brice Ménard, MD, Gaspar Rochette, S. Mallat

arXiv:2305.18512

Journal of Machine Learning Research, 25: 1-59

Export Citation

2024

A Rainbow in Deep Network Black Boxes

@article{Guth_Mallat_arxiv:2305.18512_JMLR_2024,
      title={{A Rainbow in Deep Network Black Boxes}},
      author={\text{Guth, Florentin} and M\'{e}nard, Brice and Rochette, Gaspar and \text{Mallat, St\'{{e}}phane}},
doi={10.48550/arXiv.2305.18512},
Journal = {Journal of Machine Learning Research},
Publisher= {Journal of Machine Learning Research},
abstract={A central question in deep learning is to understand the functions learned by deep networks. What is their approximation class? Do the learned weights and representations depend on initialization? Previous empirical work has evidenced that kernels defined by network acti- vations are similar across initializations. For shallow networks, this has been theoretically studied with random feature models, but an extension to deep networks has remained elu- sive. Here, we provide a deep extension of such random feature models, which we call the rainbow model. We prove that rainbow networks define deterministic (hierarchical) kernels in the infinite-width limit. The resulting functions thus belong to a data-dependent RKHS which does not depend on the weight randomness. We also verify numerically our mod- eling assumptions on deep CNNs trained on image classification tasks, and show that the trained networks approximately satisfy the rainbow hypothesis. In particular, rainbow net- works sampled from the corresponding random feature model achieve similar performance as the trained networks. Our results highlight the central role played by the covariances of network weights at each layer, which are observed to be low-rank as a result of feature learning.},
    Volume={25},
Issue={350},
year_posted={2023},
year={2024},
month={Oct},
      eprint={2305.18512},
pages={1-59},
     url= {http://jmlr.org/papers/v25/23-1573.html},
      archivePrefix={arXiv},
      Open Access Url={https://arxiv.org/abs/2305.18512}, center={CCN}, acrosscenters={yes}, external={yes}, highlight={no}
}

Download Citation (BibTex)

Vocal Call Locator Benchmark (VCL) for localizing rodent vocalizations from multi-channel audio

R. Peterson, A. Tanelus, Christopher Ick, Bartul Mimica, Niegil Francis, Violet J Ivan, A. Choudhri, A. Falkner, M. Murthy, David M Schneider, Dan Sanes, A. Williams

bioRxiv:2024.09.20.613758

The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track

Export Citation

2024

Vocal Call Locator Benchmark (VCL) for localizing rodent vocalizations from multi-channel audio

@article {Peterson_Tanelus_Williams_biorxiv:2024.09.20.613758_NEURIPS24_2024,
author = {Peterson, Ralph E and Tanelus, Aramis and Ick, Christopher and Mimica, Bartul and Francis, Niegil and Ivan, Violet J and Choudhri, Aman and Falkner, Annegret L and Murthy, Mala and Schneider, David M and Sanes, Dan H and Williams, Alex H.},
title = {{Vocal Call Locator Benchmark (VCL) for localizing rodent vocalizations from multi-channel audio}},
year = {2024},
doi = {10.1101/2024.09.20.613758},
journal = NEURIPS24,
publisher = {Cold Spring Harbor Laboratory},
abstract = {Understanding the behavioral and neural dynamics of social interactions is a goal of contemporary neuroscience. Many machine learning methods have emerged in recent years to make sense of complex video and neurophysiological data that result from these experiments. Less focus has been placed on understanding how animals process acoustic information, including social vocalizations. A critical step to bridge this gap is determining the senders and receivers of acoustic information in social interactions. While sound source localization (SSL) is a classic problem in signal processing, existing approaches are limited in their ability to localize animal-generated sounds in standard laboratory environments. Advances in deep learning methods for SSL are likely to help address these limitations, however there are currently no publicly available models, datasets, or benchmarks to systematically evaluate SSL algorithms in the domain of bioacoustics. Here, we present the VCL Benchmark: the first large-scale dataset for benchmarking SSL algorithms in rodents. We acquired synchronized video and multi-channel audio recordings of 767,295 sounds with annotated ground truth sources across 9 conditions. The dataset provides benchmarks which evaluate SSL performance on real data, simulated acoustic data, and a mixture of real and simulated data. We intend for this benchmark to facilitate knowledge transfer between the neuroscience and acoustic machine learning communities, which have had limited overlap.Competing Interest StatementThe authors have declared no competing interest.},
eprint = {https://www.biorxiv.org/content/early/2024/09/21/2024.09.20.613758.full.pdf},
URL = {https://proceedings.neurips.cc/paper_files/paper/2024/hash/c00d37d6b04d73b870b963a4d70051c1-Abstract-Datasets_and_Benchmarks_Track.html},
}

Download Citation (BibTex)

Previous Page
Viewing
Next Page