The design of handcrafted neural networks requires a lot of time and resources. Recent techniques in Neural Architecture Search (NAS) have proven to be competitive or better than traditional handcrafted design, although they require domain knowledge and have generally used limited search spaces. In this paper, we propose a novel framework for neural architecture search, utilizing a dictionary of models of base tasks and the similarity between the target task and the atoms of the dictionary; hence, generating an adaptive search space based on the base models of the dictionary. By introducing a gradient-based search algorithm, we can evaluate and discover the best architecture in the search space without fully training the networks. The experimental results show the efficacy of our proposed task-aware approach.
Federated Learning (FL) is a method of training machine learning models on private data distributed over a large number of possibly heterogeneous clients such as mobile phones and IoT devices. In this work, we propose a new federated learning framework named HeteroFL to address heterogeneous clients equipped with very different computation and communication capabilities. Our solution can enable the training of heterogeneous local models with varying computation complexities and still produce a single global inference model. For the first time, our method challenges the underlying assumption of existing work that local models have to share the same architecture as the global model. We demonstrate several strategies to enhance FL training and conduct extensive empirical evaluations, including five computation complexity levels of three model architecture on three datasets. We show that adaptively distributing subnetworks according to clients' capabilities is both computation and communication efficient.
We introduce Projected Latent Markov Chain Monte Carlo (PL-MCMC), a technique for sampling from the high-dimensional conditional distributions learned by a normalizing flow. We prove that PL-MCMC asymptotically samples from the exact conditional distributions associated with a normalizing flow. As a conditional sampling method, PL-MCMC enables Monte Carlo Expectation Maximization (MC-EM) training of normalizing flows from incomplete data. By providing experimental results for a variety of data sets, we demonstrate the practicality and effectiveness of PL-MCMC for missing data inference using normalizing flows.
Recent advances in time series classification have largely focused on methods that either employ deep learning or utilize other machine learning models for feature extraction. Though such methods have proven powerful, they can also require computationally expensive models that may lack interpretability of results, or may require larger datasets than are freely available. In this paper, we propose an interpretable baseline based on representing each time series as a collection of probability distributions of extracted geometric features. The features used are intuitive and require minimal parameter tuning. We perform an exhaustive evaluation of our baseline on a large number of real datasets, showing that simple classifiers trained on these features exhibit surprising performance relative to state of the art methods requiring much more computational power. In particular, we show that our methodology achieves good performance on a challenging dataset involving the classification of fishing vessels, where our methods achieve good performance relative to the state of the art despite only having access to approximately two percent of the dataset used in training and evaluating this state of the art.
In this paper, we demonstrate that a neural decoder trained on neural activity signals of one subject can be used to \textit{robustly} decode the motor intentions of a different subject with high reliability. This is achieved in spite of the non-stationary nature of neural activity signals and the subject-specific variations of the recording conditions. Our proposed algorithm for cross-subject mapping of neural activity is based on deep conditional generative models. We verify the results on an experimental data set in which two macaque monkeys perform memory-guided visual saccades to one of eight target locations.
It has been conjectured that the Fisher divergence is more robust to model uncertainty than the conventional Kullback-Leibler (KL) divergence. This motivates the design of a new class of robust generative auto-encoders (AE) referred to as Fisher auto-encoders. Our approach is to design Fisher AEs by minimizing the Fisher divergence between the intractable joint distribution of observed data and latent variables, with that of the postulated/modeled joint distribution. In contrast to KL-based variational AEs (VAEs), the Fisher AE can exactly quantify the distance between the true and the model-based posterior distributions. Qualitative and quantitative results are provided on both MNIST and celebA datasets demonstrating the competitive performance of Fisher AEs in terms of robustness compared to other AEs such as VAEs and Wasserstein AEs.
We present a method for learning latent stochastic differential equations (SDEs) from high dimensional time series data. Given a time series generated from a lower dimensional It\^{o} process, the proposed method uncovers the relevant parameters of the SDE through a self-supervised learning approach. Using the framework of variational autoencoders (VAEs), we consider a conditional generative model for the data based on the Euler-Maruyama approximation of SDE solutions. Furthermore, we use recent results on identifiability of semi-supervised learning to show that our model can recover not only the underlying SDE parameters, but also the original latent space, up to an isometry, in the limit of infinite data. We validate the model through a series of different simulated video processing tasks where the underlying SDE is known. Our results suggest that the proposed method effectively learns the underlying SDE, as predicted by the theory.
Rapid developments in data collecting devices and computation platforms produce an emerging number of learners and data modalities in many scientific domains. We consider the setting in which each learner holds a pair of parametric statistical model and a specific data source, with the goal of integrating information across a set of learners to enhance the prediction accuracy of a specific learner. One natural way to integrate information is to build a joint model across a set of learners that shares common parameters of interest. However, the parameter sharing patterns across a set of learners are not known a priori. Misspecifying the parameter sharing patterns and the parametric statistical model for each learner yields a biased estimator and degrades the prediction accuracy of the joint model. In this paper, we propose a novel framework for integrating information across a set of learners that is robust against model misspecification and misspecified parameter sharing patterns. The main crux is to sequentially incorporates additional learners that can enhance the prediction accuracy of an existing joint model based on a user-specified parameter sharing patterns across a set of learners, starting from a model with one learner. Theoretically, we show that the proposed method can data-adaptively select the correct parameter sharing patterns based on a user-specified parameter sharing patterns, and thus enhances the prediction accuracy of a learner. Extensive numerical studies are performed to evaluate the performance of the proposed method.