Deep Reinforcement Learning (RL) is a powerful framework for solving complex real-world problems. Large neural networks employed in the framework are traditionally associated with better generalization capabilities, but their increased size entails the drawbacks of extensive training duration, substantial hardware resources, and longer inference times. One way to tackle this problem is to prune neural networks leaving only the necessary parameters. State-of-the-art concurrent pruning techniques for imposing sparsity perform demonstrably well in applications where data distributions are fixed. However, they have not yet been substantially explored in the context of RL. We close the gap between RL and single-shot pruning techniques and present a general pruning approach to the Offline RL. We leverage a fixed dataset to prune neural networks before the start of RL training. We then run experiments varying the network sparsity level and evaluating the validity of pruning at initialization techniques in continuous control tasks. Our results show that with 95% of the network weights pruned, Offline-RL algorithms can still retain performance in the majority of our experiments. To the best of our knowledge, no prior work utilizing pruning in RL retained performance at such high levels of sparsity. Moreover, pruning at initialization techniques can be easily integrated into any existing Offline-RL algorithms without changing the learning objective.
Functional connectivity (FC) studies have demonstrated the overarching value of studying the brain and its disorders through the undirected weighted graph of fMRI correlation matrix. Most of the work with the FC, however, depends on the way the connectivity is computed, and further depends on the manual post-hoc analysis of the FC matrices. In this work we propose a deep learning architecture BrainGNN that learns the connectivity structure as part of learning to classify subjects. It simultaneously applies a graphical neural network to this learned graph and learns to select a sparse subset of brain regions important to the prediction task. We demonstrate the model's state-of-the-art classification performance on a schizophrenia fMRI dataset and demonstrate how introspection leads to disorder relevant findings. The graphs learned by the model exhibit strong class discrimination and the sparse subset of relevant regions are consistent with the schizophrenia literature.
Discovering distinct features and their relations from data can help us uncover valuable knowledge crucial for various tasks, e.g., classification. In neuroimaging, these features could help to understand, classify, and possibly prevent brain disorders. Model introspection of highly performant overparameterized deep learning (DL) models could help find these features and relations. However, to achieve high-performance level DL models require numerous labeled training samples ($n$) rarely available in many fields. This paper presents a pre-training method involving graph convolutional/neural networks (GCNs/GNNs), based on maximizing mutual information between two high-level embeddings of an input sample. Many of the recently proposed pre-training methods pre-train one of many possible networks of an architecture. Since almost every DL model is an ensemble of multiple networks, we take our high-level embeddings from two different networks of a model --a convolutional and a graph network--. The learned high-level graph latent representations help increase performance for downstream graph classification tasks and bypass the need for a high number of labeled data samples. We apply our method to a neuroimaging dataset for classifying subjects into healthy control (HC) and schizophrenia (SZ) groups. Our experiments show that the pre-trained model significantly outperforms the non-pre-trained model and requires $50\%$ less data for similar performance.
Multivariate dynamical processes can often be intuitively described by a weighted connectivity graph between components representing each individual time-series. Even a simple representation of this graph as a Pearson correlation matrix may be informative and predictive as demonstrated in the brain imaging literature. However, there is a consensus expectation that powerful graph neural networks (GNNs) should perform better in similar settings. In this work, we present a model that is considerably shallow than deep GNNs, yet outperforms them in predictive accuracy in a brain imaging application. Our model learns the autoregressive structure of individual time series and estimates directed connectivity graphs between the learned representations via a self-attention mechanism in an end-to-end fashion. The supervised training of the model as a classifier between patients and controls results in a model that generates directed connectivity graphs and highlights the components of the time-series that are predictive for each subject. We demonstrate our results on a functional neuroimaging dataset classifying schizophrenia patients and controls.
Neuroimaging studies often involve the collection of multiple data modalities. These modalities contain both shared and mutually exclusive information about the brain. This work aims at finding a scalable and interpretable method to fuse the information of multiple neuroimaging modalities using a variational autoencoder (VAE). To provide an initial assessment, this work evaluates the representations that are learned using a schizophrenia classification task. A support vector machine trained on the representations achieves an area under the curve for the classifier's receiver operating characteristic (ROC-AUC) of 0.8610.
Objective: Multimodal measurements of the same phenomena provide complementary information and highlight different perspectives, albeit each with their own limitations. A focus on a single modality may lead to incorrect inferences, which is especially important when a studied phenomenon is a disease. In this paper, we introduce a method that takes advantage of multimodal data in addressing the hypotheses of disconnectivity and dysfunction within schizophrenia (SZ). Methods: We start with estimating and visualizing links within and among extracted multimodal data features using a Gaussian graphical model (GGM). We then propose a modularity-based method that can be applied to the GGM to identify links that are associated with mental illness across a multimodal data set. Through simulation and real data, we show our approach reveals important information about disease-related network disruptions that are missed with a focus on a single modality. We use functional MRI (fMRI), diffusion MRI (dMRI), and structural MRI (sMRI) to compute the fractional amplitude of low frequency fluctuations (fALFF), fractional anisotropy (FA), and gray matter (GM) concentration maps. These three modalities are analyzed using our modularity method. Results: Our results show missing links that are only captured by the cross-modal information that may play an important role in disconnectivity between the components. Conclusion: We identified multimodal (fALFF, FA and GM) disconnectivity in the default mode network area in patients with SZ, which would not have been detectable in a single modality. Significance: The proposed approach provides an important new tool for capturing information that is distributed among multiple imaging modalities.
As evident from deep learning, very large models bring improvements in training dynamics and representation power. Yet, smaller models have benefits of energy efficiency and interpretability. To get the benefits from both ends of the spectrum we often encourage sparsity in the model. Unfortunately, most existing approaches do not have a controllable way to request a desired value of sparsity in an interpretable parameter. In this paper, we design a new sparse projection method for a set of vectors in order to achieve a desired average level of sparsity which is measured using the ratio of the $\ell_1$ and $\ell_2$ norms. Most existing methods project each vector individuality trying to achieve a target sparsity, hence the user has to choose a sparsity level for each vector (e.g., impose that all vectors have the same sparsity). Instead, we project all vectors together to achieve an average target sparsity, where the sparsity levels of the vectors is automatically tuned. We also propose a generalization of this projection using a new notion of weighted sparsity measured using the ratio of a weighted $\ell_1$ and the $\ell_2$ norms. These projections can be used in particular to sparsify the columns of a matrix, which we use to compute sparse nonnegative matrix factorization and to learn sparse deep networks.
As a mental disorder progresses, it may affect brain structure, but brain function expressed in brain dynamics is affected much earlier. Capturing the moment when brain dynamics express the disorder is crucial for early diagnosis. The traditional approach to this problem via training classifiers either proceeds from handcrafted features or requires large datasets to combat the $m>>n$ problem when a high dimensional fMRI volume only has a single label that carries learning signal. Large datasets may not be available for a study of each disorder, or rare disorder types or sub-populations may not warrant for them. In this paper, we demonstrate a self-supervised pre-training method that enables us to pre-train directly on fMRI dynamics of healthy control subjects and transfer the learning to much smaller datasets of schizophrenia. Not only we enable classification of disorder directly based on fMRI dynamics in small data but also significantly speed up the learning when possible. This is encouraging evidence of informative transfer learning across datasets and diagnostic categories.
Arguably, unsupervised learning plays a crucial role in the majority of algorithms for processing brain imaging. A recently introduced unsupervised approach Deep InfoMax (DIM) is a promising tool for exploring brain structure in a flexible non-linear way. In this paper, we investigate the use of variants of DIM in a setting of progression to Alzheimer's disease in comparison with supervised AlexNet and ResNet inspired convolutional neural networks. As a benchmark, we use a classification task between four groups: patients with stable, and progressive mild cognitive impairment (MCI), with Alzheimer's disease, and healthy controls. Our dataset is comprised of 828 subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Our experiments highlight encouraging evidence of the high potential utility of DIM in future neuroimaging studies.