Computational Knowledge Vision : The First Footprints.

Format:

Book

Author/Creator:

Zheng, Wenbo.

Contributor:

Wang, Fei-Yue.

Language:

English

Subjects (All):

Computer vision.

Information visualization.

Physical Description:

1 online resource (278 pages)

Edition:

1st ed.

Place of Publication:

San Diego : Elsevier Science & Technology, 2024.

Summary:

This book, 'Computational Knowledge Vision,' authored by Wenbo Zheng and Fei-Yue Wang, explores the integration of knowledge and vision in artificial intelligence (AI). It delves into computational frameworks that enhance machine learning and understanding of visual data, much like human perception and reasoning. By discussing various levels of vision from low to high, the book addresses edge detection, visual translation, face recognition, and visual reasoning. It also applies these concepts to practical domains such as affective computing for depression detection and medical computing for COVID-19 detection. Aimed at researchers and practitioners in AI and computer vision, the book offers insights into the evolving role of data and knowledge in advancing AI technologies. Generated by AI.

Contents:

Front Cover

Computational Knowledge Vision

Copyright

Contents

Biography

Preface

References

Acknowledgments

1 Start from here

1.1 Motivation for writing this book

1.2 Organization of the book

1 Computational Knowledge Vision framework

2 Reviewing the past enables us to learn

2.1 "Information is not knowledge"

2.2 "Knowledge is power"

2.3 Toward Computational Knowledge Vision

2.3.1 When humanoids inspire computer vision

2.3.1.1 Neural network model

2.3.1.2 Meta-learning model

2.3.2 When combinatorial mathematics meets computer vision

2.3.3 When natural language supplements computer vision

2.3.4 "Standing on the shoulders of giants"

3 Computational Knowledge Vision

3.1 What is the goal of the Computational Knowledge Vision?

3.2 Why is Computational Knowledge Vision appropriate?

3.3 What is the logic of Computational Knowledge Vision by which it can be carried out?

3.3.1 Structurized knowledge

3.3.2 Knowledge projection and conditioned feedback

3.3.3 Reasoning, understanding, and representation

2 Computational Knowledge Vision solution

4 Low vision: Computational Knowledge Vision for edge detection

4.1 Introduction

4.2 Proposed approach

4.2.1 Differential evolution algorithm

4.2.2 Differential-evolutionary-based generative adversarial networks

4.2.3 The process and structure of DEGAN

4.3 Experiments and results

4.3.1 Ablation study

4.3.2 Comparison with state-of-the-art methods

4.3.2.1 BSDS500 dataset

4.3.2.2 NYUD dataset

4.4 Conclusions

5 Middle vision: Computational Knowledge Vision for visual translation

5.1 Introduction

5.2 Just-noticeable-difference model

5.2.1 Luminance adaptation

5.2.2 Contrast masking

5.3 Proposed approach.

5.3.1 Just-noticeable-difference model of our approach

5.3.2 Network formulation

5.3.3 Network architecture

5.4 Experiments and results

5.4.1 Ablation study

5.4.2 Comparison with state-of-the-art methods

5.4.2.1 Qualitative evaluation

5.4.2.2 Quantitative evaluation

5.4.2.3 Domain adaptation

5.5 Conclusions

6 Middle vision: Computational Knowledge Vision for jointly face recognition

6.1 Introduction

6.2 Proposed approach

6.2.1 Problem definition

6.2.2 Knowledge graph construction and representation

6.2.3 Network-based representation learning

6.2.4 Knowledge-based representation learning

6.2.5 Meta-learning model

6.2.6 Meta-continual learning model

6.3 Experiments and results

6.3.1 Experimental setup

6.3.2 Comparison with the state-of-the-art methods

Sketch face recognition

Caricature face recognition

Cartoon face recognition

6.3.3 Discussion on the generalization ability

6.4 Conclusions

7 High vision: Computational Knowledge Vision for visual reasoning

7.1 Introduction

7.2 Proposed approach

7.2.1 Building knowledge bases

Multimodal Attributes Encoding

Multimodal Attributes Recovery

7.2.2 Mutual modulation model

Visual Modulation

Language Modulation

Cascaded Modulation

7.2.3 Knowledge-based key-value memory network

Key Hashing

Key Addressing and Value Reading

Hop Iterations

7.2.4 Knowledge-based representation learning

7.3 Experiments and results

7.3.1 Dataset description

7.3.2 Experiment setup

7.3.3 Comparison with state-of-the-art methods

7.3.4 Ablation study

7.3.5 Discussion about key hashing in key-value memory networks

7.4 Conclusions

3 Computational Knowledge Vision application

8 Affective computing: Computational Knowledge Vision for depression detection.

8.1 Introduction

8.2 Proposed approach

8.2.1 Basic notation

8.2.2 Multimodal attention mechanisms

8.2.3 Temporal convolution networks

8.2.4 Knowledge-based representation learning

8.2.5 Objective function

8.3 Experiments and results

8.3.1 Dataset description

DAIC-WOZ dataset

SMD dataset

8.3.2 Experiment setup

Evaluation measures

Implementation details

8.3.3 Comparison with state-of-the-art methods

Comparison on the DAIC-WOZ dataset

Comparison on the SMD dataset

8.3.4 Ablation study

8.4 Conclusions

9 Medical computing: Computational Knowledge Vision for COVID-19 detection

9.1 Introduction

9.2 Multimodal dataset construction

9.2.1 Dataset creation and structure

9.2.2 Dataset comparison

9.3 Proposed approach

9.3.1 Problem definition

9.3.2 Network representation learning

9.3.3 Data augmentation

9.3.4 Self-knowledge distillation

9.3.5 Training methods

9.4 Experiments and results

9.4.1 Experimental settings

9.4.2 The results of our model

9.4.3 Comparison with state-of-the-art methods

COVID-19 diagnosis from pneumonia cases on our dataset

COVID-19 cases diagnosis on COVID v2.0 dataset

9.5 Conclusions

10 Medical computing: Computational Knowledge Vision for medical visual reasoning

10.1 Introduction

10.2 Proposed approach

10.2.1 Problem setup

10.2.2 Knowledge graph construction and representation

10.2.3 Feature fusion network

10.2.4 Knowledge-based representation learning

10.2.5 Meta-learning model

10.3 Experiments and results

10.3.1 Dataset description

10.3.2 Experiment setup

10.3.3 Comparison with state-of-the-art methods

10.3.4 Discussion about other similar methods

Summary about comparison and discussion experiments

10.3.5 Error analysis

10.4 Conclusions

References.

Index

Back Cover.

Notes:

Description based on publisher supplied metadata and other sources.

Part of the metadata in this record was created by AI, based on the text of the resource.

Other Format:

Print version: Zheng, Wenbo Computational Knowledge Vision

ISBN:

9780443216183

OCLC:

1453197564

Publisher Number:

CIPO000110012