In the virtual hand interaction techniques, the opacity of the virtual hand avatar can potentially obstruct users' visual feedback, leading to detrimental effects on accuracy and cognitive load. Given that the cog...
详细信息
ISBN:
(纸本)9798350374490;9798350374506
In the virtual hand interaction techniques, the opacity of the virtual hand avatar can potentially obstruct users' visual feedback, leading to detrimental effects on accuracy and cognitive load. Given that the cognitive load is related to gaze movements, our study focuses on analyzing the gaze movements of participants across opaque, transparent, and invisible hand visualizations in order to create a new interaction technique. For our experimental setup, we used a Purdue Pegboard Test with reaching, grasping, transporting, and inserting subtasks. We examined how long and where participants concentrated on these subtasks and, using the findings, introduced a new virtual hand visualization method to increase accuracy. We hope that our results can be used in future virtual reality applications where users have to interact with virtual objects accurately.
Prompts, which are pieces of text injected into the inputs of NLP models, can significantly enhance these models' performance. Despite their effectiveness, a straightforward method for exploring a large number of ...
详细信息
ISBN:
(纸本)9798350393811;9798350393804
Prompts, which are pieces of text injected into the inputs of NLP models, can significantly enhance these models' performance. Despite their effectiveness, a straightforward method for exploring a large number of possible prompts and assessing their quality in downstream tasks is lacking. This paper introduces PromptLandscape, a visualization approach to exploring prompts in the zero-shot image classification context. Our focus is on the CLIP model [26], which has been pre-trained on a large number of image-text pairs and exhibits exceptional performance in zero-shot image classifications. However, its performance is also considerably influenced by the text labels. Prompts were introduced to refine the text labels and improve the classification performance. Nevertheless, the prompts are often proposed through trial-and-error and it is challenging to find the best prompts for different image datasets. PromptLandscape presents the large number of possible prompts as points in a 3D space to guide their exploration and enables users to decode any points back as texts for better interpretation. Through concrete case studies, we demonstrate the effectiveness of PromptLandscape in (1) finding the best prompts for image classifications and (2) interpreting the prompts to reveal the crucial features of the studied image datasets.
As data becomes more and more pervasive in our daily lives, supporting people in getting visual insights into their data is an important challenge to address. However, as data visualization literacy is still low, a ga...
详细信息
ISBN:
(纸本)9798350341614
As data becomes more and more pervasive in our daily lives, supporting people in getting visual insights into their data is an important challenge to address. However, as data visualization literacy is still low, a gap between designers' mental model and users is not uncommon. To best pick up users where they are, we propose to include the data mental models of users into the design process. In this paper, we present our investigations in this direction by incorporating user sketches of their idea about their personal data stored at online services as a basis of our designs. We present our design study on personal data visualization interfaces resulting in a set of user sketches for three types of user groups and in three visualization interfaces. Finally, we reflect on our learnings and identify pitfalls to support other researchers in applying similar approaches.
We propose a visual query system for risky enterprises. Based on a small number of input samples of risky enterprises, users can interactively analyze and obtain SQL statements that express the query results, facilita...
详细信息
ISBN:
(纸本)9798350393811;9798350393804
We propose a visual query system for risky enterprises. Based on a small number of input samples of risky enterprises, users can interactively analyze and obtain SQL statements that express the query results, facilitating subsequent task operations and understanding risk criteria. Due to the diversity of enterprises and the complexity of economic indicators, understanding, evaluating, and querying risky enterprises is a challenging task for ordinary users. To solve this problem, our visual query system integrates multicriteria decision-making techniques and data mining through a visual analysis interface. The system facilitates efficient identification of risky enterprises, leveraging knowledge gained from a small number of known risk company samples. Our system includes an analytic hierarchy model to express domain knowledge, a query tree construction algorithm, and a well-designed visual interface. Two real cases demonstrate the effectiveness of our system.
The rapid development of generative AI (GenAI) models in computer vision necessitates effective evaluation methods to ensure their quality and fairness. Existing tools primarily focus on dataset quality assurance and ...
详细信息
ISBN:
(纸本)9798350393811;9798350393804
The rapid development of generative AI (GenAI) models in computer vision necessitates effective evaluation methods to ensure their quality and fairness. Existing tools primarily focus on dataset quality assurance and model explainability, leaving a significant gap in GenAI output evaluation during model development. Current practices often depend on developers' subjective visual assessments, which may lack scalability and generalizability. This paper bridges this gap by conducting a formative study with GenAI model developers in an industrial setting. Our findings led to the development of GenLens, a visual analytic interface designed for the systematic evaluation of GenAI model outputs during the early stages of model development. GenLens offers a quantifiable approach for overviewing and annotating failure cases, customizing issue tags and classifications, and aggregating annotations from multiple users to enhance collaboration. A user study with model developers reveals that GenLens effectively enhances their workflow, evidenced by high satisfaction rates and a strong intent to integrate it into their practices. This research underscores the importance of robust early-stage evaluation tools in GenAI development, contributing to the advancement of fair and high-quality GenAI models.
Virtual reality (VR) lets users explore objects in an immersive environment. Users freely interact with objects in a virtual environment (VE), where they can assign physical properties to objects, make them animate or...
详细信息
ISBN:
(纸本)9798350374490;9798350374506
Virtual reality (VR) lets users explore objects in an immersive environment. Users freely interact with objects in a virtual environment (VE), where they can assign physical properties to objects, make them animate or inanimate, and have them respond according to the underlying characteristics. This research explores geological topology using VR. In addition, users studied a terrain model in a multimodal audio-visual VE. Initial findings showed that users preferred multimodal exploration of terrain models compared to the traditional unimodal vision-only version. Two audio-enhanced modes were used: "Continuous" and "Sporadic."
In a feedforward network, Transfer Entropy (TE) can be used to measure the influence that one layer has on another by quantifying the information transfer between them during training. According to the information Bot...
详细信息
ISBN:
(纸本)9798350341614
In a feedforward network, Transfer Entropy (TE) can be used to measure the influence that one layer has on another by quantifying the information transfer between them during training. According to the information Bottleneck principle, a neural model's internal representation should compress the input data as much as possible while still retaining sufficient information about the output. information Plane analysis is a visualization technique used to understand the trade-off between compression and information preservation in the context of the information Bottleneck method by plotting the amount of information in the input data against the compressed representation. The claim that there is a causal link between information-theoretic compression and generalization, measured by mutual information, is plausible, but results from different studies are conflicting. In contrast to mutual information, TE can capture temporal relationships between variables. To explore such links, in our novel approach we use TE to quantify information transfer between neural layers and perform information Plane analysis. We obtained encouraging experimental results, opening the possibility for further investigations.
The cluster faithfulness metrics CQ measure how faithfully the ground truth clustering of a graph is represented as the geometric clustering in a drawing of the graph. Existing CQ metrics use k-means clustering, which...
详细信息
ISBN:
(纸本)9798350393811;9798350393804
The cluster faithfulness metrics CQ measure how faithfully the ground truth clustering of a graph is represented as the geometric clustering in a drawing of the graph. Existing CQ metrics use k-means clustering, which effectively compute a geometric clustering when the cluster sizes are even, resulting in accurate CQ metrics. However, k-means clustering tends to compute clusters of even sizes and thus often fails to compute an accurate geometric clustering when the cluster sizes are uneven, leading to inaccurate CQ metrics. In this paper, we present a new cluster faithfulness metric CQHAC, using HAC (Hierarchical Agglomerative Clustering). HAC can compute a more accurate geometric clustering for uneven cluster sizes than k-means clustering. Consequently, CQ-HAC can more accurately measure cluster faithfulness, regardless of whether the sizes of clusters are even or uneven. Moreover, we present two algorithms, Cluster-kmeans and Cluster-HAC, for optimizing cluster faithfulness of graph drawings. Extensive experiments show that in practice, both algorithms always compute perfectly clusterfaithful drawings (i.e., CQ = 1) in our experiments using various graphs with both even and uneven cluster sizes, achieving significant improvement over existing graph layouts, including cluster-focused layouts.
We introduce CLeVer, a novel visualization system designed to analyze and enhance the performance of continual learning models by detecting task transition failures. Based on the literature review, we discovered and c...
详细信息
ISBN:
(纸本)9798350393811;9798350393804
We introduce CLeVer, a novel visualization system designed to analyze and enhance the performance of continual learning models by detecting task transition failures. Based on the literature review, we discovered and classified three primary causes of task transition failure in classification tasks: class/position inconsistency, biased/noisy samples, and diverse class scopes. These problems are critical in terms of model performance but were not tackled in prior research on continual learning. CLeVer is designed to address these challenges as an integrated system for model experiments and visual analysis. Firstly, users can easily configure the tasks for continual learning through a single JSON file. Then, our system automatically simulates the continual learning process in a relatively short time while generating data for visualization. Finally, the transition visualizer provides an effective visual representation of the task transition process where users can easily detect the task transition failures in continual learning. Our interview with machine learning experts and a case study with three participants demonstrate CLeVer's utility in detecting and addressing such failures. We also discuss the system's potential applicability and adaptability for various computer vision tasks while suggesting our future work.
Virtual self-avatars have been shown to produce the Proteus effect, however limited work investigates the subjective sense of embodiment using commodity virtual reality systems. In this work, we present results from a...
详细信息
ISBN:
(纸本)9798350374490;9798350374506
Virtual self-avatars have been shown to produce the Proteus effect, however limited work investigates the subjective sense of embodiment using commodity virtual reality systems. In this work, we present results from a pilot experiment where participants are given a self-avatar in a simple virtual experience while wearing a cardboard head-mounted display. Participants then repeat the experience five days later. Overall, subjective embodiment scores are similar to those reported in experience using higher-fidelity systems. However, the subjective sense of embodiment significantly lowered from trial one to trial two.
暂无评论