Image by editor
# introduction
When I first started learning AI, I spent a lot of time copying code from tutorials, only to realize that I didn’t really understand how AI worked. The real skill is not just running the model. It’s about knowing why they work and how to apply them to real problems. AI books helped me learn concepts, reasoning, and practical aspects of AI in a way that simple tutorials could not. With that in mind, we are proposing this series. Free but really valuable book. This article is aimed at anyone who wants to learn AI, and our initial recommendations are listed below.
# 1. Neural networks and deep learning
book Neural networks and deep learning Learn the basics of neural networks to actually building and training deep models yourself. Starting with simple ideas like perceptrons and sigmoid neurons, we walk you through creating a network that can recognize handwritten digits. You’ll also see how backpropagation actually works to train these models, and how to improve them using cost functions, regularization, weight initialization, hyperparameter tuning, and more. There are many Python code samples so you can test it yourself and see how everything connects. It’s a good combination of both intuition and mathematics, so it’s not just about understanding; how Neural networks work, but why. If you already know some math (linear algebra, calculus, etc.), this is a good choice to not only use the library, but to really see what’s going on under the hood.
// Summary overview:
- Fundamentals of neural networks (Perceptron, sigmoid neuron, network architecture, handwritten digit classification, gradient descent, network implementation)
- Backpropagation and learning (matrix-based computation, cost function assumptions, Hadamard product, four basic backpropagation equations, algorithm implementation, learning improvements)
- advanced training techniques (Cross-entropy cost, overfitting and regularization, weight initialization, hyperparameter selection, universality of neural nets, extending beyond sigmoid neurons)
- Deep learning and convolutional networks (Vanishing gradient problem, unstable gradient, convolutional neural network, practical application, recent progress in image recognition, future direction)
# 2. Deep learning
deep learning It provides a very good overview of deep learning and how machines actually learn from experience and build complex ideas from simple ones. You’ll start with the necessary math pieces, such as linear algebra, probability, information theory, and a bit of number crunching, and then learn the basics of machine learning. We then dive deeper into modern deep learning techniques such as feedforward, convolutional networks, recurrent networks, regularization, and optimization, and show how they are used in real-world projects. We also cover advanced topics such as autoencoders, generative and representation learning, and structured probabilistic models. It is primarily written for people with a solid mathematical knowledge, so it is more of a reference for research and advanced work than a guide for beginners.
// Summary overview:
- Factor models and autoencoders (PCA, ICA, sparse coding, incomplete and regularized autoencoders, denoising, manifold learning)
- Representation learning and probabilistic models (Stratified pre-learning, transfer learning, distributed representation, structured probabilistic model, approximate inference, Monte Carlo method)
- Deep generative models and advanced techniques (Boltzmann machine, deep belief network, convolutional model, generative stochastic network, autoencoder sampling, generative model evaluation)
# 3. Practical deep learning
link:
free course Practical deep learning It’s designed for people who already know some coding and want to get hands-on with machine learning and deep learning. Instead of just reading theory, you can immediately start building models for real-world tasks. This course covers modern tools such as Python. pie torch,and fast tie It shows you how to use the library to handle everything from data cleaning to model training, testing, and deployment. Learn by doing with real notebooks, datasets, and problems. It focuses on practical, modern methods for selecting the right algorithms, properly validating, scaling, and deploying them.
// Summary overview:
- Fundamentals and model training (Neural network fundamentals, stochastic gradient descent, affine functions and nonlinearity, backpropagation, MLP, autoencoders)
- Cross-domain applications (Computer vision using CNN, natural language processing (NLP) including embeddings and phrase similarity, tabular data modeling, collaborative filtering and recommendations)
- Advanced technology and optimization (Transfer learning, weight decay, data augmentation, accelerated stochastic gradient descent (SGD), ResNets, mixed precision, DDPM/DDIM, attention and transformers, latent diffusion, super resolution)
- Introduction and practical skills (Converting models to web applications, improving accuracy/speed/reliability, ethical considerations, frameworks such as The Learner, matrix operations, model initialization/normalization)
# 4. Artificial Intelligence: Fundamentals of Computational Agents
book Artificial Intelligence: Fundamentals of Computational Agents describes AI through the concept of a “computational agent,” a system that can sense, learn, reason, and act. The latest edition adds new topics such as neural networks, deep learning, causality, and social and ethical aspects of AI. It shows how agents are constructed, how they plan and act, and how they deal with complex or uncertain situations. Each chapter contains an algorithm. pythoncase studies, and real-life discussions to learn both the how and the why. A well-balanced mix of theory and practice, it’s perfect for students and anyone looking for a modern and in-depth introduction to AI.
// Summary overview:
- Fundamentals of AI and agents (Examples include natural and artificial intelligence, historical context, agent design spaces, delivery robots, diagnostic assistants, tutoring systems, transactional agents, and smart homes).
- Agent architecture and control (Hierarchical control, agent functionality, offline and online computation, how agents perceive and act within their environments.)
- Reasoning, planning, searching (Problem solving using search, graph traversal, constraint satisfaction, probabilistic reasoning, and planning techniques such as forward planning, regression planning, and partial order programming)
- Learning and neural networks (Supervised learning, decision trees, complex models such as regression, overfitting, boosting, deep learning architectures (convolutional neural networks (CNNs), recurrent neural networks (RNNs), transformers), large-scale language models).
- Uncertainty, causality, and reinforcement learning (Probabilistic reasoning, Bayesian learning, unsupervised methods, causal inference, decision making under uncertainty, sequential decision making, reinforcement learning strategies such as Q-learning and evolutionary algorithms).
# 5. Ethical artificial intelligence
paper ethical artificial intelligence We consider how future AI systems may behave in unexpected or potentially harmful ways and suggest ways to design them safely. First, they point out that AI can learn models of the world that are far more complex than humans can fully understand, making safeguards difficult. The authors recommend using utility functions (mathematical descriptions of what the AI should care about) rather than vague rules, as the goals are clearer. We also address issues such as self-delusion, where AIs can corrupt their own observations and rewards, unintentional “shortcuts” that can harm us, and corrupted reward generators, where AIs manipulate their own reward systems. The authors propose a model that learns human values, uses finite definitions, and incorporates self-modeling to allow AI to reason about its own actions. We also consider the bigger picture, including how AI will impact society, politics, and the future of humanity.
// Summary overview:
- Fundamentals and AI design (Future AI and current AI, instructional AI, utility maximizing agent, learning environment model, intelligence measures, ethical framework)
- AI behavior and challenges (self-delusion, unintended instrumental actions, model-based utility functions, learning human values, evolved embedded agents)
- Testing, governance and society (AI testing, real-world behavior, political aspects, transparency, benefit sharing, ethical considerations)
- Philosophical and social influence (Search for meaning, social and cultural influences, bridging computation and human values)
# summary
These books (and articles, and courses) cover a wide range of what AI engineers need, from neural networks and deep learning to hands-on coding, agent-based AI, and ethical issues. They provide a clear path from learning ideas to applying AI in real-world situations. What topic would you like us to cover next? Leave your suggestions in the comments section.
kanwar mereen I’m a machine learning engineer and technical writer with a deep passion for the intersection of data science, AI, and healthcare. She co-authored the e-book “Maximize Productivity with ChatGPT.” She champions diversity and academic excellence as a 2022 Google Generation Scholar for APAC. She has also been recognized as a Teradata Diversity in Tech Scholar, Mitacs Globalink Research Scholar, and a Harvard WeCode Scholar. Kanwal is a passionate advocate for change and founded FEMCodes to empower women in STEM fields.
