Share E-Book
Scan to open this page

Scan with your phone to open this page

Author: Ömer Faruk Ertuğrul, Tahir Çetin Akinci, Musa Yilmaz

Rating 3.5 AI initial rating

Understanding Models Developed by AI: Including Applications with Python and MATLAB Code is a comprehensive guide on the intricacies of AI models and their real-world applications. The book demystifies complex AI methodologies by providing clear explanations and practical examples that are reinforced with Python and MATLAB program codes. Its content structure emphasizes a practical, applications-driven approach to understanding AI models, with hands-on coding examples throughout each chapter. Readers will find the tools they need to build AI models, along with the knowledge to make these models accessible and interpretable to stakeholders, thus fostering trust and reliability in AI systems. As the primary issues with the adoption of AI/ML models are reliability, transparency, interpretation of results, and bias (data and algorithm) management, this resource give researchers and developers what they need to be able to not only implement AI models, but also interpret and explain them. This is crucial in industries where decision-making processes must be transparent and understandable.

AI Reading Assistant

Whole-book reading guide from stratified index samples; jump to passages in the text

AI guide
【One-Line Pitch】 A practical, code-backed guide to building AI models that you can actually explain—covering everything from feature selection and error metrics to deep learning and ethics—for graduate students, engineers, and data scientists who need transparency, not just accuracy. 【Book Arc】 - **Opening (~0%–15%)**: Sets the stakes—why explainability matters for trust, fairness, and adoption—then introduces XAI terminology, a taxonomy (scope, timing, model relationship, specificity), and a review of core model families from regression and decision trees to deep learning and LLMs. - **Early (~15%–33%)**: Moves into hands-on fundamentals: model explanation techniques (intrinsic vs. post-hoc, global vs. local, surrogate models, LIME/SHAP, saliency), feature selection and data augmentation with Python/MATLAB examples, and interpreting performance and error metrics. - **Middle (~33%–56%)**: Applies interpretation to specific model families—classification (with Grad-CAM analysis), regression (LIME, Shapley values, sensitivity analysis), clustering and pattern discovery, and time-series/decision problems—each paired with code and visual explanation. - **Late (~56%–80%)**: Tackles the "opaque" methods head-on: deep learning architectures (CNNs, RNN/LSTM, transformers) with attribution and visualization techniques, plus reinforcement learning including a healthcare scenario and its security/ethical dimensions. - **Ending (~80%–100%)**: Closes with applied case studies (e.g., ANN-based HHV prediction from biomass parameters) and a chapter on AI ethics and responsible use—bias, privacy, accountability, federated learning, and generative AI—tying interpretability to regulation and trust. 【Key Takeaways】 - **Explainability is a design requirement, not an afterthought** (Opening): The book frames reliability, transparency, bias management, and interpretability as the core barriers to AI adoption, making XAI the organizing theme rather than a bonus chapter. - **A working taxonomy separates explanation types** (Early): Scope (global/local), timing (pre-model/in-model/post-model), and model-specificity give readers a vocabulary to choose the right explanation tool for a given model and audience. - **Data decisions drive interpretability** (Early): Feature selection and data augmentation are presented as levers that affect both predictive performance and how explainable the resulting model is—with runnable Python and MATLAB examples. - **Metrics must be read, not just computed** (Early): A dedicated treatment of error metrics and evaluation pipelines in both languages teaches readers to interpret what a number actually says about model behavior. - **Model-family chapters pair theory with code and visuals** (Middle): Classification, regression, clustering, and time-series each get a concise theoretical overview followed by implementation and interpretation via figures, tables, and case studies. - **Even "black-box" methods can be pried open** (Late): Deep learning and reinforcement learning chapters apply LIME, SHAP, Integrated Gradients, Grad-CAM, and saliency maps to reveal what the model is actually using to decide. - **Ethics and interpretability are linked** (Ending): Fairness, accountability, privacy, and emerging regulation are treated as consequences of explainability choices, not separate concerns. - **Balance accuracy with clarity** (Throughout): The recurring thesis is that interpretable models need not be trivial and high-performing systems can still be examined, tested, and audited. 【Reading Tips】 - **Skim Chapter 1 if you already know XAI basics**, but keep its taxonomy handy—later chapters repeatedly map techniques back to scope, timing, and specificity. - **Deep-read the code sections** (feature selection, metrics, classification/regression interpretation) with Python or MATLAB open; the book's value is in reproducible examples, not prose alone. - **Treat model-family chapters as modular**: jump to the one matching your work (e.g., deep learning or regression) rather than reading linearly. - **Don't skip the ethics chapter** even if you're purely technical—it reframes interpretability as an accountability and regulatory issue. - **Watch for the recurring chapter template** (problem → design → code → interpretation), which makes it easy to extract the takeaway from each chapter quickly. 【Coverage Limits】 Excerpts are dominated by front matter, table of contents, and preface, so chapter-level detail beyond titles and stated scope is thin; specific results, datasets, and code specifics in later chapters are not covered here.
Excerpt 1
Models Developed with AI This page intentionally left blank Understanding Models Developed with AI Including Applications with Python and MATLAB Code Edited...
View in text
Page 7
3.2 General principles of data augmentation techniques 85 3.3 Data augmentation in different application areas 87 3.4 Application example: Simple data augmen...
View in text
Excerpt 3
hical Approaches 298 References 299 Index 303 viii Contents Contributors Mustafa Ilhan Akbas, Embry-Riddle Aeronautical Uni- versity, Electrical Engineering...
View in text
Page 14
ubiquitous, integrated into various applications and fields. Advanced AI and machine learning (ML) models, particularly deep learning (DL), large language mo...
View in text
Page 15
r instance, the debt-to-income ratio has the highest weight. Other common terms used in XAI include transparency and fairness. Transparency refers to the cla...
View in text
Page 16
dels use simple, predefined if-then rules to make decisions. Owing to the transparent relationship between inputs and outputs, rule-based models are inherent...
View in text
Page 18
data types beyond text, making them suitable for many tasks. Furthermore, with the development of agentic AI, LLMs can perform a variety of tasks with autono...
View in text
Page 20
relationships in the data and boost confidence in the model. For example, understanding the important features that affect credit application results, such a...
View in text
Tags
AI categories
Artificial IntelligencePythonMatlab
ISBN: 0443441642
Publisher: Morgan Kaufmann
Publish Year: 2026
Language: English
Pages: 322
File Format: PDF
File Size: 23.8 MB
Text Preview (First 20 pages)
Registered users can read the full content for free

Register as a Gaohf Library member to read the complete e-book online for free and enjoy a better reading experience.

Generating text preview…