Ai Academy : Deep Learning

Fixed the material of the content, the material are more up to date for 2025 and all information got better and more up to date :  Now is 80 articles fixed and updated for 2025 to learn ai - deep learning step by step in 80 days : topics covered are : 80 days topics include in this app are : Intro to ML – ML types, models, train-test split. ML in iOS – Using ML in apps. Model Types – Supervised vs. unsupervised. Regression & Classification – Basics with MNIST. SGD Math – How it works. Normal Equation – Prediction without iteration. Gradient Descent – Concept and application. Types of GD – Batch, Stochastic, Mini-Batch. Perceptrons – Deep learning basics. MLPs – Regression vs. classification. Activation Functions – ReLU, Sigmoid, etc. Non-Linearity – Hidden layers & activation. Intro to Keras – Building models. Keras APIs – Sequential, Functional, Subclassing. Keras API Comparison – Sequential vs. Functional. TensorFlow Tools – TensorBoard, callbacks, saving models. Hyperparameter Tuning – With Keras Tuner. Manual vs. Auto Optimization – Tuning models. Bayesian Optimization – Neural network tuning. Vanishing Gradient – Explanation. Weight Initialization – Strategies. Monetizing AI APIs – Creating paid APIs. Weight Init: Part 2 – Advanced methods. Advanced Activation – GELU, Mish, etc. Batch Norm – How it works. Batch Norm: Part 2 – Further details. Batch Norm Parameters – Trainable/non-trainable. Gradient Clipping – Avoid exploding gradients. Transfer Learning – Basics. Transfer Learning Example – Implementing it. Labeled vs. Unlabeled Data – Differences. Speeding Up Training – Optimization tricks. Momentum Optimization – Deep dive. Momentum vs. Normalization – Comparisons. Momentum: Part 3 – Advanced details. NAG Optimizer – Nesterov Accelerated Gradient. AdaGrad – Origins & proof. Optimizer Comparison – AdaGrad, RMSProp, Adam. Adam vs. Other Optimizers – Pros & cons. Adam & Local Minima – Understanding behavior. More Optimizers – NAdam, AdaMax, AdamW. Learning Rate Schedules – Why they matter. 1Cycle Schedule – Explanation. Gradient Clipping & Weight Init – Combined effect. LR Scheduling Methods – 1-Cycle, CED, Exponential. TensorFlow vs. PyTorch vs. MLX – Frameworks compared. Regularization – Preventing overfitting. Dropout – Including MC Dropout. Max-Norm Regularization – Explanation. Deep vs. Dense Networks – Differences. Deep Learning Use Cases – Overview. ML in iOS Apps – Integration. CNNs – Basics & use cases. CNN Math – How it works. RNNs: Part 1 – Sequence modeling. RNNs: Part 2 – More details. RNNs & Time Series – Forecasting. RNN vs. Feedforward – Mathematical comparison. ARIMA & SARIMA – Before diving into RNNs. RNN Step-by-Step – Time series forecasting. Seq2Seq Forecasting – Iterative vs. direct. LSTMs & Layer Norm – RNN enhancements. RNNs for NLP – Language modeling. Why Transformers Win in NLP – Key insights. Transformers Overview – GPT to DeepSeek. Transformer Breakthroughs – ChatGPT to DeepSeek. BERT Explained – Key insights. How ChatGPT Works – Basics. ChatGPT vs. BERT – Understanding comparison. ChatGPT: Step-by-Step – Breakdown. NLP Mathematics – Behind modern AI models. Transformers in Vision & Multimodal AI – Expansion. Autoencoders, GANs, & Diffusion – Overview. Stacked Autoencoders – Unsupervised pretraining. Diffusion Models – Breaking them down. GANs – Deep learning for image generation. How DALL·E Works – Image synthesis. Reinforcement Learning – Applications & impact. DeepNet – Scaling Transformers to 1,000 layers. DeepSeek-R1 – Advancing LLM reasoning.