Treatment Learning Causal Transformer for Noisy Image Classification

Yang, Chao-Han Huck; Hung, Danny I-Te; Liu, Yi-Chieh; Chen, Pin-Yu

doi:10.1109/wacv56688.2023.00608

Cited by 8 publications

(3 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most current methods, like those in references [72,73], look at patterns between image features and labels. However, these methods can be thrown off by random patterns [61] or when data isn't consistent [60]. That's why it's crucial to use causal theory, which helps understand the direct cause-and-effect between images and labels.…”

Section: Practical: Problem Formulationmentioning

confidence: 99%

Computer Vision with Causal Inference/Learning: A Deep Learning Approach Notes

Hambarde

2023

Preprint

View full text Add to dashboard Cite

Deep learning heavily relies on statistical correlations to drive artificial intelligence (AI) innovations, particularly in computer vision applications like autonomous driving and robotics. However, despite providing a solid foundation for deep learning, these statistical correlations can be vulnerable to unforeseen and uncontrolled factors. The lack of prior knowledge guidance can result in spurious correlations, introducing confounding factors and affecting the model's robustness. To address this challenge, recent research efforts have focused on integrating causal theory into deep learning methodologies. By modelling the inherent and unbiased causal structure, causal theory can potentially mitigate the impact of spurious correlations effectively. Hence, this paper explores the basics of causal methodologies in image classification.

show abstract

Section: Practical: Problem Formulationmentioning

confidence: 99%

Computer Vision with Causal Inference/Learning: A Deep Learning Approach Notes

Hambarde

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In general, causal learning for real-world observational studies is complicated [1,2,7,[19][20][21][22][23][24][25][28][29][30][31][32][33][34]44,45,59,[61][62][63]. With the established efforts [45,49,66] on causal learning under noisy interventions, two assumptions are imposed when modeling the problem of video summarization.…”

Section: Assumptionsmentioning

confidence: 99%

Causal Video Summarizer for Video Exploration

Huang

Yang

Chen³

et al. 2022

2022 IEEE International Conference on Multimedia and Expo (ICME)

View full text Add to dashboard Cite

The goal of video summarization is to automatically shorten videos such that it conveys the overall story without losing relevant information. In many application scenarios, improper video summarization can have a large impact. For example in forensics, the quality of the generated video summary will affect an investigator's judgment while in journalism it might yield undesired bias. Because of this, modeling explainability is a key concern. One of the best ways to address the explainability challenge is to uncover the causal relations that steer the process and lead to the result. Current machine learning-based video summarization algorithms learn optimal parameters but do not uncover causal relationships. Hence, they suffer from a relative lack of explainability. In this work, a Causal Explainer, dubbed Causalainer, is proposed to address this issue. Multiple meaningful random variables and their joint distributions are introduced to characterize the behaviors of key components in the problem of video summarization. In addition, helper distributions are introduced to enhance the effectiveness of model training. In visual-textual input scenarios, the extra input can decrease the model performance. A causal semantics extractor is designed to tackle this issue by effectively distilling the mutual information from the visual and textual inputs. Experimental results on commonly used benchmarks demonstrate that the proposed method achieves state-of-the-art performance while being more explainable.

show abstract

“…Still, on distribution network transformers, Hu et al (2021) present a study using characteristics of the transformer failure rate in its life cycle to perform an opportunity cost analysis considering the maintenance and replacement of equipment. Yang et al (2023) present a 10kV distribution transformer replacement investment prediction model based on Lasso and GBDT algorithms. The model is used to forecast investments in replacing transformers in distribution networks in China.…”

Section: Introductionmentioning

confidence: 99%

Asset Management Support Tool for Energy Systems using AHP and Monte Carlo Methods applied to Power Transformers

Campanhola,

Schaefer,

Carraro

et al. 2023

IJEEP

View full text Add to dashboard Cite

The assets of electric power transmission systems are characterized by their high cost and complexity, leading utilities to seek ways to improve the efficiency and economic-financial performance of these assets. Thus, this article proposes a tool to support asset management for electricity transmission systems. This tool considers the useful life of the equipment, its importance to the system, financial aspects, and the forecast of electricity demand. An evaluation of the criteria that impact decision-making on the replacement of assets with specialists was carried out. Thus, the tool uses the Analytic Hierarchy Process method to classify the most critical equipment in the system. For analysis of future scenarios, the Monte Carlo Method was incorporated into the tool to simulate the behavior of equipment during a defined time horizon. As a result, the tool presents a ranking of the most critical equipment in the system under analysis within the simulated time horizon. The tool was applied in a case study with real data in the power transformer of a Brazilian utility. The tool helps in decision-making indicating changes that may be made in the period under review, their likely impact on equipment loading, and the list of critical transformers in the system.

show abstract

Treatment Learning Causal Transformer for Noisy Image Classification

Cited by 8 publications

References 61 publications

Computer Vision with Causal Inference/Learning: A Deep Learning Approach Notes

Computer Vision with Causal Inference/Learning: A Deep Learning Approach Notes

Causal Video Summarizer for Video Exploration

Asset Management Support Tool for Energy Systems using AHP and Monte Carlo Methods applied to Power Transformers

Contact Info

Product

Resources

About