Nous utilisons des témoins pour analyser le trafic et l’utilisation de notre site web, afin de personnaliser votre expérience. Vous pouvez désactiver ces technologies à tout moment, mais cela peut restreindre certaines fonctionnalités du site. Consultez notre Politique de protection de la vie privée pour en savoir plus.
Paramètre des cookies
Vous pouvez activer et désactiver les types de cookies que vous souhaitez accepter. Cependant certains choix que vous ferez pourraient affecter les services proposés sur nos sites (ex : suggestions, annonces personnalisées, etc.).
Cookies essentiels
Ces cookies sont nécessaires au fonctionnement du site et ne peuvent être désactivés. (Toujours actif)
Cookies analyse
Acceptez-vous l'utilisation de cookies pour mesurer l'audience de nos sites ?
Multimedia Player
Acceptez-vous l'utilisation de cookies pour afficher et vous permettre de regarder les contenus vidéo hébergés par nos partenaires (YouTube, etc.) ?
Publications
Not one model fits all: unfairness in RSFC-based prediction of behavioral data in African American
14 Helmholtz AI kick-off meeting 5 Mar 2020, 14:17:33 Page 1/1 Abstract #14 | Poster Not one model fits all: unfairness in RSFC-based predic… (voir plus)tion of behavioral data in African American J. Li , D. Bzdok, A. Holmes, T. Yeo, S. Genon 1 Forschungszentrum Julich, Institute of Neuroscience and Medicine, Jülich, Germany 2 McGill University, Department of Biomedical Imaging, Montreal, Canada 3 National University of Singapore, ECE, CSC, CIRC, N.1 & MNP, Singapore, Singapore 4 Yale University, New Haven, United States of America While predictive models are expected to play a major role in personalized medicine approaches in the future, biases towards specific population groups have been evidenced, hence raising concerns about the risks of unfairness of machine learning algorithms. As great hopes and intense work have been invested recently in the prediction of behavioral phenotypes based on brain resting-state functional connectivity (RSFC), we here examined potential differences in RSFC-based predictive models of behavioral data between African American (AA) and White American (WA) samples matched for the main demographic, anthropometric, behavioral and in-scanner motion variables. We used resting-fMRI data with 58 behavioral measures of 953 subjects comprising 130 African American (AA) and 724 White American (WA). For each subject, a 419 x 419 matrix summarizing connectivity of 419 brain regions was computed. Matching between AA and WA was performed at the subject level by creating 102 pairs of AA and WA subjects, matched for 6 types of variables (age, sex, intracranial volume, education, in-scanner motion and behavioral scores). We performed 10-fold nested cross-validation by randomly splitting the 102 pairs across 10 sets. The remaining 749 subjects were also divided across the 10 sets. A predictive model was built for each behavioral variable by using kernel ridge regression. All analyses focused on the 102 matched AA and WA groups. After FDR correction (q 0.05), no significant difference was found between the matched AA and WA groups for the matching variables. Out of 58 behavioral variables, 38 showed significantly above chance prediction accuracies (based on permutation test, FDR corrected). Overall, average prediction performance for these variables was higher in the WA group than in the AA group. Furthermore, significant differences in prediction performance between the two groups were found in 35 behavioral variables (FDR corrected; q 0.05). Our results suggest that RSFC-based prediction models of behavioral phenotype trained on the entire HCP population show different prediction performance in different subsets of the population. This suggest that one model might not fit all that, in some cases, RSFC-based predictive models might have poorer prediction accuracies for African Americans compared to matched White Americans. Future work should evaluate the factors contributing to these discrepancies and the potential consequences, as well as possible recommendations.
RandomNet: Towards Fully Automatic Neural Architecture Design for Multimodal Learning
Almost all neural architecture search methods are evaluated in terms of performance (i.e. test accuracy) of the model structures that it fin… (voir plus)ds. Should it be the only metric for a good autoML approach? To examine aspects beyond performance, we propose a set of criteria aimed at evaluating the core of autoML problem: the amount of human intervention required to deploy these methods into real world scenarios. Based on our proposed evaluation checklist, we study the effectiveness of a random search strategy for fully automated multimodal neural architecture search. Compared to traditional methods that rely on manually crafted feature extractors, our method selects each modality from a large search space with minimal human supervision. We show that our proposed random search strategy performs close to the state of the art on the AV-MNIST dataset while meeting the desirable characteristics for a fully automated design process.
The tensor network formalism has enjoyed over two decades of success in modeling the behavior of complex quantum-mechanical systems, but has… (voir plus) only recently and sporadically been leveraged in machine learning. Here we introduce a uniform matrix product state (u-MPS) model for probabilistic modeling of sequence data. We identify several distinctive features of this recurrent generative model, notably the ability to condition or marginalize sampling on characters at arbitrary locations within a sequence, with no need for approximate sampling methods. Despite the sequential architecture of u-MPS, we show that a recursive evaluation algorithm can be used to parallelize its inference and training, with a string of length n only requiring parallel time
Tensor networks are a powerful modeling framework developed for computational many-body physics, which have only recently been applied withi… (voir plus)n machine learning. In this work we utilize a uniform matrix product state (u-MPS) model for probabilistic modeling of sequence data. We first show that u-MPS enable sequence-level parallelism, with length-n sequences able to be evaluated in depth O(log n). We then introduce a novel generative algorithm giving trained u-MPS the ability to efficiently sample from a wide variety of conditional distributions, each one defined by a regular expression. Special cases of this algorithm correspond to autoregressive and fill-in-the-blank sampling, but more complex regular expressions permit the generation of richly structured text in a manner that has no direct analogue in current generative models. Experiments on synthetic text data find u-MPS outperforming LSTM baselines in several sampling tasks, and demonstrate strong generalization in the presence of limited data.
2020-03-02
International Conference on Artificial Intelligence and Statistics (publié)
Tensor networks are a powerful modeling framework developed for computational many-body physics, which have only recently been applied withi… (voir plus)n machine learning. In this work we utilize a uniform matrix product state (u-MPS) model for probabilistic modeling of sequence data. We first show that u-MPS enable sequence-level parallelism, with length-n sequences able to be evaluated in depth O(log n). We then introduce a novel generative algorithm giving trained u-MPS the ability to efficiently sample from a wide variety of conditional distributions, each one defined by a regular expression. Special cases of this algorithm correspond to autoregressive and fill-in-the-blank sampling, but more complex regular expressions permit the generation of richly structured text in a manner that has no direct analogue in current generative models. Experiments on synthetic text data find u-MPS outperforming LSTM baselines in several sampling tasks, and demonstrate strong generalization in the presence of limited data.
2020-03-02
International Conference on Artificial Intelligence and Statistics (published)
Electric load forecasting, especially short-term load forecasting (STLF), is becoming more and more important for power system operation. We… (voir plus) propose to use multiple kernel learning (MKL) for residential electric load forecasting which provides more flexibility than traditional kernel methods. Computation time is an important issue for short-term forecasting, especially for energy scheduling. However, conventional MKL methods usually lead to complicated optimization problems. Another practical issue for this application is that there may be a very limited amount of data available to train a reliable forecasting model for a new house, while at the same time we may have historical data collected from other houses which can be leveraged to improve the prediction performance for the new house. In this paper, we propose a boosting-based framework for MKL regression to deal with the aforementioned issues for STLF. In particular, we first adopt boosting to learn an ensemble of multiple kernel regressors and then extend this framework to the context of transfer learning. Furthermore, we consider two different settings: homogeneous transfer learning and heterogeneous transfer learning. Experimental results on residential data sets demonstrate that forecasting error can be reduced by a large margin with the knowledge learned from other houses.
Model-free deep reinforcement learning is sample inefficient. One hypothesis -- speculated, but not confirmed -- is that catastrophic interf… (voir plus)erence within an environment inhibits learning. We test this hypothesis through a large-scale empirical study in the Arcade Learning Environment (ALE) and, indeed, find supporting evidence. We show that interference causes performance to plateau; the network cannot train on segments beyond the plateau without degrading the policy used to reach there. By synthetically controlling for interference, we demonstrate performance boosts across architectures, learning algorithms and environments. A more refined analysis shows that learning one segment of a game often increases prediction errors elsewhere. Our study provides a clear empirical link between catastrophic interference and sample efficiency in reinforcement learning.
Normalizing flows are powerful invertible probabilistic models that can be used to translate two probability distributions, in a way that al… (voir plus)lows us to efficiently track the change of probability density. However, to trade for computational efficiency in sampling and in evaluating the log-density, special parameterization designs have been proposed at the cost of representational expressiveness. In this work, we propose to use ODEs as a framework to establish universal approximation theory for certain families of flow-based models.
2020-02-26
International Conference on Learning Representations (publié)