An official website of the United States government
Traditional unsupervised domain adaptation methods attempt to align source and target domains globally and are agnostic to the categories of the data points. This results in an inaccurate categorical alignment and diminishes the classification performance on the target domain. In this paper, we alter existing adversarial domain alignment methods to adhere to category alignment by imputing category information. We partition the samples based on category using source labels and target pseudo labels and then apply domain alignment for every category. Our proposed modification provides a boost in performance even with a modest pseudo label estimator. We evaluate our approach on 4 popular domain alignment loss functions using object recognition and digit datasets.
more »
« less
How I Learned What a Domain Was
This paper re-traverses the author's investigations across several years as he sought to pin-down the meaning of the in vivo category 'domain'. The paper is a methodological reflection on the grounded theory approach to concept development, with a focus on the technical terms: in vivo category, iteration on the code, and sensitizing category. It is also a substantive theoretical contribution, elaborating the concept of a domain in computing, data and information science, and how it has long served as an organizing principle for developing computational systems. Four tricks of the trade for studying the 'logic of domains' are offered as sensitizing concepts to aid future investigations.
more »
« less
- Award ID(s):
- 1638903
- PAR ID:
- 10602025
- Publisher / Repository:
- Association for Computing Machinery (ACM)
- Date Published:
- Journal Name:
- Proceedings of the ACM on Human-Computer Interaction
- Volume:
- 3
- Issue:
- CSCW
- ISSN:
- 2573-0142
- Format(s):
- Medium: X Size: p. 1-12
- Size(s):
- p. 1-12
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
The concept of hidden genes was recently introduced in genetic algorithms (GAs) to handle systems architecture optimization problems, where the number of design variables is variable. Selecting the hidden genes in a chromosome determines the architecture of the solution. This paper presents two categories of mechanisms for selecting (assigning) the hidden genes in the chromosomes of GAs. These mechanisms dictate how the chromosome evolves in the presence of hidden genes. In the proposed mechanisms, a tag is assigned for each gene; this tag determines whether the gene is hidden or not. In the first category of mechanisms, the tags evolve using stochastic operations. Eight different variations in this category are proposed and compared through numerical testing. The second category introduces logical operations for tags evolution. Both categories are tested on the problem of interplanetary trajectory optimization for a space mission to Jupiter, as well as on mathematical optimization problems. Several numerical experiments were designed and conducted to optimize the selection of the hidden genes algorithm parameters. The numerical results presented in this paper demonstrate that the proposed concept of tags and the assignment mechanisms enable the hidden genes genetic algorithms (HGGA) to find better solutions.more » « less
-
Topological domain walls separating 2+1 dimensional topologically ordered phases can be understood in terms of Witt equivalences between the UMTCs describing anyons in the bulk topological orders. However, this picture does not provide a framework for decomposing stacks of multiple domain walls into superselection sectors — i.e., into fundamental domain wall types that cannot be mixed by any local operators. Such a decomposition can be understood using an alternate framework in the case that the topological order is anomaly-free, in the sense that it can be realized by a commuting projector lattice model. By placing these Witt equivalences in the context of a 3-category of potentially anomalous (2+1)D topological orders, we develop a framework for computing the decomposition of parallel topological domain walls into indecomposable superselection sectors, extending the previous understanding to topological orders with non-trivial anomaly. We characterize the superselection sectors in terms of domain wall particle mobility, which we formalize in terms of tunnelling operators. The mathematical model for the 3-category of topological orders is the 3-category of fusion categories enriched over a fixed unitary modular tensor category.more » « less
-
Reliable failure detection holds paramount importance in safety-critical applications.Yet, neural networks are known to produce overconfident predictions for misclassified samples. As a result, it remains a problematic matter as existing confidence score functions rely on category-level signals, the logits, to detect failures. This research introduces an innovative strategy, leveraging human-level concepts for a dual purpose: to reliably detect when a model fails and to transparently interpret why.By integrating a nuanced array of signals for each category, our method enables a finer-grained assessment of the model's confidence.We present a simple yet highly effective approach based on the ordinal ranking of concept activation to the input image. Without bells and whistles, our method is able to significantly reduce the false positive rate across diverse real-world image classification benchmarks, specifically by 3.7% on ImageNet and 9.0% on EuroSAT.more » « less
-
The parathyroid cell is a vital regulator of extracellular calcium levels, operating through the secretion of parathyroid hormone (PTH). Despite its importance, the regulation of PTH secretion remains complex and not fully understood, representing a unique interplay between extracellular and intracellular calcium, and hormone secretion. One significant challenge in parathyroid research has been the difficulty in maintaining cells ex vivo for in-depth cellular investigations. To address this issue, we introduce a novel platform for parathyroid cell transplantation and noninvasive in vivo imaging using the anterior chamber of the eye as a transplantation site. We found that parathyroid adenoma tissue transplanted into the mouse eye engrafted onto the iris, became vascularized, and retained cellular composition. Transplanted animals exhibited elevated PTH levels, indicating a functional graft. With in vivo confocal microscopy, we were able to repetitively monitor parathyroid graft morphology and vascularization. In summary, there is a pressing need for new methods to study complex cellular processes in parathyroid cells. Our study provides a novel approach for noninvasive in vivo investigations that can be applied to understand parathyroid physiology and pathology under physiological and pathological conditions. This innovative strategy can deepen our knowledge on parathyroid function and disease.more » « less
