skip to main content

Search for: All records

Award ID contains: 1845587

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Electroencephalogram (EEG) signals are often used as an input modality for Brain Computer Interfaces (BCIs). While EEG signals can be beneficial for numerous types of interaction scenarios in the real world, high levels of noise limits their usage to strictly noise-controlled environments such as a research laboratory. Even in a controlled environment, EEG is susceptible to noise, particularly from user motion, making it highly challenging to use EEG, and consequently BCI, as a ubiquitous user interaction modality. In this work, we address the EEG noise/artifact correction problem. Our goal is to detect physiological artifacts in EEG signal and automatically replacemore »the detected artifacts with imputed values to enable robust EEG sensing overall requiring significantly reduced manual effort than is usual. We present a novel EEG state-based imputation model built upon a recurrent neural network, which we call SRI-EEG, and evaluate the proposed method on three publicly available EEG datasets. From quantitative and qualitative comparisons with six conventional and neural network based approaches, we demonstrate that our method achieves comparable performance to the state-of-the-art methods on the EEG artifact correction task.« less
    Free, publicly-accessible full text available May 20, 2023
  2. Imperfect labels are ubiquitous in real-world datasets. Several recent successful methods for training deep neural networks (DNNs) robust to label noise have used two primary techniques: filtering samples based on loss during a warm-up phase to curate an initial set of cleanly labeled samples, and using the output of a network as a pseudo-label for subsequent loss calculations. In this paper, we evaluate different augmentation strategies for algorithms tackling the "learning with noisy labels" problem. We propose and examine multiple augmentation strategies and evaluate them using synthetic datasets based on CIFAR-10 and CIFAR-100, as well as on the real-world datasetmore »Clothing1M. Due to several commonalities in these algorithms, we find that using one set of augmentations for loss modeling tasks and another set for learning is the most effective, improving results on the state-of-the-art and other previous methods. Furthermore, we find that applying augmentation during the warm-up period can negatively impact the loss convergence behavior of correctly versus incorrectly labeled samples. We introduce this augmentation strategy to the state-of-the-art technique and demonstrate that we can improve performance across all evaluated noise levels. In particular, we improve accuracy on the CIFAR-10 benchmark at 90% symmetric noise by more than 15% in absolute accuracy, and we also improve performance on the Clothing1M dataset.« less
  3. Modern machine learning algorithms typically require large amounts of labeled training data to fit a reliable model. To minimize the cost of data collection, researchers often employ techniques such as crowdsourcing and web scraping. However, web data and human annotations are known to exhibit high margins of error, resulting in sizable amounts of incorrect labels. Poorly labeled training data can cause models to overfit to the noise distribution, crippling performance in real-world applications. In this work, we investigate the viability of using data augmentation in conjunction with semi-supervised learning to improve the label noise robustness of image classification models. Wemore »conduct several experiments using noisy variants of the CIFAR-10 image classification dataset to benchmark our method against existing algorithms. Experimental results show that our augmentative SSL approach improves upon the state-of-the-art.« less
  4. Many researchers and industry professionals believe Augmented Reality (AR) to be the next step in personal computing. However, the idea of an always-on context-aware AR device presents new and unique challenges to the way users organize multiple streams of information. What does multitasking look like and when should applications be tied to specific elements in the environment? In this exploratory study, we look at one such element: physical objects, and explore an object-centric approach to multitasking in AR. We developed 3 prototype applications that operate on a subset of objects in a simulated test environment. We performed a pilot studymore »of our multitasking solution with a novice user, domain expert, and system expert to develop insights into the future of AR application design.« less
  5. Curating large and high quality datasets for studying affect is a costly and time consuming process, especially when the labels are continuous. In this paper, we examine the potential to use unlabeled public reactions in the form of textual comments to aid in classifying video affect. We examine two popular datasets used for affect recognition and mine public reactions for these videos. We learn a representation of these reactions by using the video ratings as a weakly supervised signal. We show that our model can learn a fine-grained prediction of comment affect when given a video alone. Furthermore, we demonstratemore »how predicting the affective properties of a comment can be a potentially useful modality to use in multimodal affect modeling.« less
  6. Augmented reality (AR) interfaces increasingly utilize artificial intelligence systems to tailor content and experiences to the user. We explore the effects of one such system — a recommender system for online shopping — which allows customers to view personalized product recommendations in the physical spaces where they might be used. We describe results of a [Formula: see text] condition exploratory study in which recommendation quality was varied across three user interface types. Our results highlight potential differences in user perception of the recommended objects in an AR environment. Specifically, users rate product recommendations significantly higher in AR and in amore »3D browser interface, and show a significant increase in trust in the recommender system, compared to a web interface with 2D product images. Through semi-structured interviews, we gather participant feedback which suggests AR interfaces perform better due to their ability to view products within the physical context where they will be used.« less