Variational Autoencoder with Truncated Mixture of Gaussians for Functional Connectivity Analysis

Zhao, Qingyu; Honnorat, Nicolas; Adeli, Ehsan; Pfefferbaum, Adolf; Sullivan, Edith V.; Pohl, Kilian M.

doi:10.1007/978-3-030-20351-1_68

Cited by 21 publications

(17 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…deep networks has often been defined by minimizing the reconstruction error [34], such as in auto-encoders (AEs). AEs have shown to be great tools for unsupervised representation learning in a variety of tasks, including image inpainting [43], feature ranking [54], denosing [57], clustering [65], defense against adversarial examples [35], and anomaly detection [48,52]. Although AEs have led to farreaching success for data representation, there are some caveats associated with using reconstruction errors as the sole metric for representation learning: (1) As also argued in [58], it forces to reconstruct all parts of the input, even if they are irrelevant for any given task or are contaminated by noise; (2) It leads to a mechanism that entirely depends on single-point data abstraction, i.e., the AE learns to just reconstruct its input while neglecting other data points present in the dataset.…”

Section: Introductionmentioning

confidence: 99%

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Sabokrou

Khalooei

Adeli

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

In this paper, we propose a novel self-supervised representation learning by taking advantage of a neighborhoodrelational encoding (NRE) among the training data. Conventional unsupervised learning methods only focused on training deep networks to understand the primitive characteristics of the visual data, mainly to be able to reconstruct the data from a latent space. They often neglected the relation among the samples, which can serve as an important metric for self-supervision. Different from the previous work, NRE aims at preserving the local neighborhood structure on the data manifold. Therefore, it is less sensitive to outliers. We integrate our NRE component with an encoder-decoder structure for learning to represent samples considering their local neighborhood information. Such discriminative and unsupervised representation learning scheme is adaptable to different computer vision tasks due to its independence from intense annotation requirements. We evaluate our proposed method for different tasks, including classification, detection, and segmentation based on the learned latent representations. In addition, we adopt the auto-encoding capability of our proposed method for applications like defense against adversarial example attacks and video anomaly detection. Results confirm the performance of our method is better or at least comparable with the state-of-the-art for each specific application, but with a generic and self-supervised approach.

show abstract

Section: Introductionmentioning

confidence: 99%

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Sabokrou

Khalooei

Adeli

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Modeling the temporal dynamics is desirable but non-trivial, since it is highly irregular, complex and variable. To fill this gap, we direct future studies to designing a recurrent neural network (Chen and Hu, 2018;Cui et al, 2019;Shi et al, 2018;Sutskever et al, 2014;Zhao et al, 2019), as an add-on to VAE, to further learn sequence representation, for example, with a self-supervised predictive learning strategy (Kashyap and Keilholz, 2020;Khosla et al, 2019a).…”

Section: Discussionmentioning

confidence: 99%

Representation Learning of Resting State fMRI with Variational Autoencoder

Kim

Zhang

Han

et al. 2020

Preprint

View full text Add to dashboard Cite

20Resting state functional magnetic resonance imaging (rs-fMRI) data exhibits complex 21 but structured patterns. However, the underlying origins are unclear and entangled in rs-22 fMRI data. Here we establish a variational auto-encoder, as a generative model 23 trainable with unsupervised learning, to disentangle the unknown sources of rs-fMRI 24 activity. After being trained with large data from the Human Connectome Project, the 25 model has learned to represent and generate patterns of cortical activity and 26 connectivity using latent variables. Of the latent representation, its distribution reveals 27 overlapping functional networks, and its geometry is unique to each individual. Our 28 results support the functional opposition between the default mode network and the 29 task-positive network, while such opposition is asymmetric and non-stationary. 30Correlations between latent variables, rather than cortical connectivity, can be used as a 31 more reliable feature to accurately identify subjects from a large group, even if only a 32 short period of data is available per subject. 33 34 3

show abstract

“…VAEs have recently been popularized for non-rs-fMRI data and gained attention due to their interpretable latent space and ability to variationally learn generative factors that fit a certain prior. Previous work evaluates representation learning with VAEs on rs-fMRI data that has first undergone dimensionality reduction [8]- [11]. These dimensionality reductions may incur overly specific inductive biases and, as a result, limit the expressivity of deep learning methods, especially since neural networks are considered universal function approximators.…”

Section: A Contextmentioning

confidence: 99%

Variational voxelwise rs-fMRI representation learning: Evaluation of sex, age, and neuropsychiatric signatures

Geenjaar¹,

White²,

Calhoun³

2021

Preprint

View full text Add to dashboard Cite

We propose to apply non-linear representation learning to voxelwise rs-fMRI data. Learning the non-linear representations is done using a variational autoencoder (VAE). The goal is to use a VAE trained on voxelwise rs-fMRI data to perform non-linear dimensionality reduction that retains meaningful information. The retention of information in the model's representations is evaluated using downstream age regression and sex classification tasks for the UK Biobank dataset. The results on these tasks are highly encouraging and a linear regressor trained with the representations of our unsupervised model performs almost as well as a supervised neural network, trained specifically for age regression on the same dataset. Another important quality of our model is that the representations with which age regression and sex classification are performed are the same. This implies that the representations can retain information about both the sex classification and age regression tasks, something a supervised model would have to specifically be trained for with multi-task learning and signifies its potential as a dimensionality reduction method.The model is also evaluated with a schizophrenia diagnosis prediction task, to assess its feasibility as a dimensionality reduction method for neuropsychiatric datasets. These results highlight the potential for pre-training on a larger set of individuals who do not have mental illness, to improve the downstream neuropsychiatric task results. The pre-trained model is fine-tuned for a variable number of epochs on a schizophrenia dataset and we find that fine-tuning for 1 epoch yields the best results. This work therefore not only opens up non-linear dimensionality reduction for voxelwise rs-fMRI data but also shows that pretraining a deep learning model on voxelwise rs-fMRI datasets greatly increases performance even on smaller datasets. It also opens up the ability to look at the distribution of rs-fMRI time series in the latent space of the VAE for heterogeneous neuropsychiatric disorders like schizophrenia in future work. This can be complemented with the generative aspect of the model that allows us to reconstruct points from the model's latent space back into brain space and obtain an improved understanding of the relation that the VAE learns between subjects, timepoints, and a subject's characteristics.

show abstract

Variational Autoencoder with Truncated Mixture of Gaussians for Functional Connectivity Analysis

Cited by 21 publications

References 16 publications

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Representation Learning of Resting State fMRI with Variational Autoencoder

Variational voxelwise rs-fMRI representation learning: Evaluation of sex, age, and neuropsychiatric signatures

Contact Info

Product

Resources

About