We deliver AI algorithm tutorials and the latest news, curated by a team of researchers from Peking University, Shanghai Jiao Tong University, Central South University, and leading AI companies such as Huawei, Kuaishou, and SenseTime. Join us in the AI alchemy—making life better!
This article explains DeepMind’s PonderNet framework, which lets any neural network allocate computation adaptively, demonstrates its implementation with PyTorch Lightning on the MNIST dataset, details the underlying theory, loss functions, training procedure, and evaluates its pondering behavior on rotated digit experiments.
This article derives the policy‑gradient objective for discrete actions, implements the Monte‑Carlo REINFORCE algorithm in PyTorch, explains the actor‑critic framework, introduces Advantage Actor‑Critic (A2C) versus A3C, and demonstrates their performance on the OpenAI Gym CartPole‑v0 environment.
This article explains the principal component analysis (PCA) method for reducing data dimensions, walks through the mathematical steps, shows how to implement it manually with NumPy, and demonstrates reuse with scikit‑learn's PCA class, including concrete code examples and output.
This article explains the SMOTE oversampling method for imbalanced classification, demonstrates how to generate synthetic minority samples, evaluates models with and without SMOTE using scikit‑learn pipelines, and explores advanced variants such as Borderline‑SMOTE, SVMSMOTE and ADASYN with concrete code examples and benchmark results.
This article walks through building a PyTorch ShuffleNetV2 model, preparing the Kaggle Flowers dataset, training with transfer learning on a GPU, visualizing loss and accuracy, and performing inference on five test images, achieving nearly 90% validation accuracy after 95 epochs.
This article walks through creating a satellite image classification pipeline using PyTorch and a pretrained ResNet34 model, covering dataset preparation, project structure, data loading, model definition, training, validation, loss/accuracy plotting, and inference on new images with detailed code examples and results.
This article provides a step‑by‑step guide to building, training, and evaluating a custom object detection model using PyTorch Faster R‑CNN on a microcontroller dataset, covering data preparation, configuration, model modification, training loops, loss visualization, and inference on new images.
This article explains the purpose and types of feature selection, compares supervised and unsupervised, wrapper, filter, and embedded methods, discusses choosing statistical metrics based on variable types, and provides scikit‑learn code examples for regression and classification tasks.
The article explains how high‑dimensional data causes the curse of dimensionality, reduces model performance, and surveys feature‑selection, matrix‑decomposition, manifold‑learning, and auto‑encoder techniques while advising systematic experiments and proper data scaling.
This article explains how to modify the Soft Actor‑Critic (SAC) algorithm—originally designed for continuous actions—to work with discrete action environments, presents the required changes to the actor and critic loss functions, provides a full PyTorch implementation, and evaluates the method on the CartPole‑v1 benchmark.
