Recent works showed that Generative Adversarial Networks (GANs) can be successfully applied in unsupervised domain adaptation, where, given a labeled source dataset and an unlabeled target dataset, the goal is to train powerful classifiers for the target samples. In particular, it was shown that a GAN objective function can be used to learn target features indistinguishable from the source ones. In this work, we extend this framework by (i) forcing the learned feature extractor to be domain-invariant, and (ii) training it through data augmentation in the feature space, namely performing feature augmentation. While data augmentation in the image space is a well established technique in deep learning, feature augmentation has not yet received the same level of attention. We accomplish it by means of a feature generator trained by playing the GAN minimax game against source features. Results show that both enforcing domain-invariance and performing feature augmentation lead to superior or comparable performance to state-of-the-art results in several unsupervised domain adaptation benchmarks.
The potential of conjugated polymers to compete with inorganic materials in the field of semiconductor is conditional on fine-tuning of the charge carriers mobility. The latter is closely related to the material morphology, and various studies have shown that the bottleneck for charge transport is the connectivity between well-ordered crystallites, with a high degree of π -π stacking, dispersed into a disordered matrix. However, at this time there is a lack of theoretical descriptions accounting for this link between morphology and mobility, hindering the development of systematic material designs. Here we propose a computational model to predict charge carriers mobility in conducting polymer PEDOT depending on the physicochemical properties of the system. We start by calculating the morphology using molecular dynamics simulations. Based on the calculated morphology we perform quantum mechanical calculation of the transfer integrals between states in polymer chains and calculate corresponding hopping rates using the Miller-Abrahams formalism. We then construct a transport resistive network, calculate the mobility using a mean-field approach, and analyze the calculated mobility in terms of transfer integrals distributions and percolation thresholds. Our results provide theoretical support for the recent study [Noriega et al., Nat. Mater. 12, 1038] explaining why the mobility in polymers rapidly increases as the chain length is increased and then saturates for sufficiently long chains. Our study also provides the answer to the long-standing question whether the enhancement of the crystallinity is the key to designing high-mobility polymers. We demonstrate, that it is the effective π -π stacking, not the long-range order that is essential for the material design for the enhanced electrical performance. This generic model can compare the mobility of a polymer thin film with different solvent contents, solvent additives, dopant species or polymer characteristics, providing a general framework to design new high mobility conjugated polymer materials.
We are concerned with the vulnerability of computer vision models to distributional shifts. We formulate a combinatorial optimization problem that allows evaluating the regions in the image space where a given model is more vulnerable, in terms of image transformations applied to the input, and face it with standard search algorithms. We further embed this idea in a training procedure, where we define new data augmentation rules according to the image transformations that the current model is most vulnerable to, over iterations. An empirical evaluation on classification and semantic segmentation problems suggests that the devised algorithm allows to train models that are more robust against content-preserving image manipulations and, in general, against distributional shifts 1 . * VM is also with Huawei Technologies (Ireland) Co., Ltd., Dublin. 1 Code at github.com/ricvolpi/domain-shift-robustness
Dropout is a very effective way of regularizing neural networks. Stochastically "dropping out" units with a certain probability discourages over-specific co-adaptations of feature detectors, preventing overfitting and improving network generalization. Besides, Dropout can be interpreted as an approximate model aggregation technique, where an exponential number of smaller networks are averaged in order to get a more powerful ensemble. In this paper, we show that using a fixed dropout probability during training is a suboptimal choice. We thus propose a time scheduling for the probability of retaining neurons in the network. This induces an adaptive regularization scheme that smoothly increases the difficulty of the optimization problem. This idea of "starting easy" and adaptively increasing the difficulty of the learning problem has its roots in curriculum learning and allows one to train better models. Indeed, we prove that our optimization strategy implements a very general curriculum scheme, by gradually adding noise to both the input and intermediate feature representations within the network architecture. Experiments on seven image classification datasets and different network architectures show that our method, named Curriculum Dropout, frequently yields to better generalization and, at worst, performs just as well as the standard Dropout method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.