Tag: 机器学习
62 articles
AI Research #135: Gemini 3 Pro Back on Top - MoE, Million...
Explains Gemini 3 Pro's advantages through sparse MoE architecture, million-token context, native multimodal (text/image/video/PDF), thinking depth control (thinking_level), and Deep Think mode. St...
AI Research #132: Java Ecosystem 2025 - Spring, Quarkus, ...
Spring Framework 6, Spring Boot 3.x (minimum JDK 17, supports Java 21 Virtual Threads), GraalVM Native Image AOT, Quarkus and Micronaut cloud-native...
AI Research #130: Qwen2.5-Omni Practical Applications
Office assistant, education and training, programming and operations, search-enhanced RAG, device control/plugin agents, and companion entertainment. Covers...
AI Research #129: Qwen2.5-Omni-7B Key Specs - VRAM, Conte...
Runs stably at FP16 ~14GB VRAM, with INT8/INT4 quantization (<4GB) enabling deployment on consumer GPUs or edge devices. Combined with FlashAttention 2 and...
AI Research #128: Qwen2.5-Omni Training Pipeline - Three-...
Complete training pipeline breakdown for Qwen2.5-Omni: Thinker based on Qwen2.5, vision initialized from Qwen2.5-VL, audio from Whisper-large-v3. Uses...
AI Research #127: Qwen2.5-Omni Deep Dive - Thinker-Talker...
Engineering breakdown of Qwen2.5-Omni (2024-2025) Thinker-Talker dual-core architecture: unified Transformer decoder for text/image/video/audio fusion, TMRoPE...
AI Research #125: Tesla FSD Business Model and Competitor...
FSD V14 (2025) business model and competitive landscape. Analyzes pricing logic for one-time purchase (~15,000) vs subscription (~199/month) and deferred...
AI Research #124: Tesla FSD V14 Deep Analysis
Tesla FSD V14 real-world performance and road tests, comparing V13.2 on urban roads and highways: key disengagement metrics, lane changes/ramps, destination arrival, and long-tail scenarios. V14 sh...
AI Research #123: FSD V14 Deep Analysis - Vision-Only SDF...
FSD V14 (2025) technical evolution compared to V12 (2023), focusing on vision-only approach, SDF (Signed Distance Field) occupancy reconstruction, end-to-end...
AI Research #121: DeepSeek-OCR Research Directions
Frontier approaches and engineering implementation for DeepSeek-OCR (2025, including 3B parameter direction). Summarizes research directions including...
AI Research #119: DeepSeek-OCR PyTorch FlashAttn 2.7.3 In...
Comprehensive guide for DeepSeek-OCR local/private deployment based on Python 3.12, PyTorch 2.6.0, Transformers 4.46.3 and FlashAttention 2.7.3. Includes ~3B parameter model inference, deployment o...
AI Research #120: DeepSeek-OCR from 0 to 1 - Getting Star...
Complete getting started path and engineering essentials for DeepSeek-OCR (as of 2025), covering environment setup (Python/PyTorch 2.x, Transformers 4.x), model loading, output parsing, parameter e...
AI Research #118: Embodied AI Mobile-ALOHA - Mobile + Dua...
Mobile-ALOHA: An open-source mobile manipulation solution combining mobile chassis and dual-arm collaboration. Uses whole-body teleoperation for low-cost...
AI Investigation #108: Complete Robot Model Training Proc...
Full robot training pipeline: pre-training, fine-tuning (LoRA), reinforcement learning, imitation learning, and human feedback for safe autonomous decision-making.
AI Investigation #107: RL and Robot Training Data Format ...
Data formats and development processes in robot and reinforcement learning systems, including time series trajectories, state-action pairs, offline RL data,...
AI Investigation #106: Robot Learning Data Collection Too...
Core data collection methods and application scenarios, covering over ten methods from manual entry, sensor collection, web crawlers, API calls, log collection...
AI Investigation #105: Robot Learning Data Collection - F...
Data collection is a critical step in robot learning development, covering demonstration video collection, trajectory recording, state-action pair generation, and data quality control strategies.
AI Investigation #103: Embodied AI Technology Landscape
Comprehensive overview of embodied AI tech stack: hardware (GPU, sensors, actuators), software (ROS, simulation), and algorithms (deep learning, RL, VLA models).
AI Investigation #102: Intelligent Robotic Arms, Autonomo...
Different types of robots have huge differences in structure, tasks and control methods, so AI algorithm adaptation strategies also need to be tailored.
AI Investigation #101: Modern AI Methods - VLA, RT-1, RT-...
Modern AI robot control methods are undergoing a major transition from reinforcement learning and imitation learning to multimodal agents driven by large models. The combination of Vision-Language-...
AI Investigation #100: Modern AI Methods - Reinforcement ...
Modern AI methods for robot control cover Reinforcement Learning (RL), Imitation Learning (IL), and Transformer-based large model methods. Reinforcement...
AI Investigation #99: Sensor Fusion Technology - Camera, ...
Sensor Fusion is a core technology in autonomous driving, robotics and smart security. Through multi-sensor data fusion of cameras, LiDAR, radar, IMU,...
AI Investigation #98: Visual SLAM - ORB-SLAM, RTAB-Map, V...
Visual SLAM is a technology that achieves autonomous positioning and environment mapping without relying on LiDAR, using only cameras. By extracting environmental features (corners, edges, textures...
AI Investigation #97: SLAM Algorithm Comparison and Appli...
Multi-sensor fusion and SLAM are core technologies for robot perception and navigation. By fusing IMU, GPS, wheel odometry, LiDAR, visual odometry and other...
AI Investigation #96: Robot Scenario Testing - From Extre...
Complete guide to robot scenario testing, covering three dimensions: environment testing, load testing, and anomaly testing. Traditional manual testing has...
AI Investigation #95: Robot Scenario Testing - From Extre...
Before robots enter practical applications, systematic scenario testing must be conducted, covering boundary conditions like extreme weather, complex terrain,...
AI Investigation #93: Robot Simulation Tools - Comprehens...
Simulation tools are an important part of robot R&D, enabling algorithm verification and system debugging in risk-free environments, accelerating iteration.
AI Investigation #92: Robot Motion Control - From Traditi...
Robot motion control can be divided into two categories: traditional model-based methods and deep learning-based intelligent control. The former emphasizes kinematics/dynamics modeling, trajectory ...
AI Investigation #91: Multi-modal Data Annotation Tools -...
In robot vision and perception model training, high-quality multi-modal data annotation tools are crucial. Current mainstream solutions cover 2D images, video, text, audio and 3D point cloud multi-...
Spark MLlib GBDT Case Study: Residual Calculation to Regr...
GBDT practical case study walking through the complete process from residual calculation to regression tree construction and iterative training. Covers GBDT...
Spark MLlib: Bagging vs Boosting Differences and GBDT Gra...
Introduces the differences between Bagging and Boosting in machine learning, and the GBDT (Gradient Boosting Decision Tree) algorithm principles. Main content:...
Spark MLlib GBDT Algorithm: Gradient Boosting Principles,...
This article introduces the principles and applications of gradient boosting tree (GBDT) algorithm. First explains boosting tree basic concept through simple examples, then details algorithm flow i...
Spark MLlib Ensemble Learning: Random Forest, Bagging and...
This article systematically introduces ensemble learning methods in machine learning. Main content includes: 1) Basic definition and classification of ensemble...
Spark MLlib Decision Tree Pruning: Pre-pruning, Post-prun...
This article systematically introduces decision tree pre-pruning and post-pruning principles, compares core differences between three mainstream algorithms...
Spark MLlib Decision Tree: Classification Principles, Gin...
This article introduces the basic concepts, classification principles, and classification principles of decision trees. Decision tree is a non-linear...
Spark MLlib Logistic Regression: Input Function, Sigmoid,...
This article introduces the basic principles, application scenarios, and implementation in Spark MLlib of logistic regression. Logistic regression is an efficient binary classification algorithm wi...
Spark MLlib Linear Regression: Scenarios, Loss Function a...
Linear regression uses regression equations to model relationships between independent and dependent variables. This article covers regression scenarios (house...
Spark MLlib Logistic Regression: Sigmoid, Loss Function a...
Logistic regression is a classification model in machine learning — an efficient binary classification algorithm widely used in ad click-through rate...
Spark MLlib Linear Regression: Scenarios, Loss Function a...
Linear Regression is an analytical method that uses regression equations to model the relationship between one or more independent variables and a dependent...
sklearn KMeans Key Attributes & Evaluation: cluster_cente...
scikit-learn (sklearn) KMeans (2026) explains three most commonly used objects: cluster_centers_ (cluster centers), inertia_ (Within-Cluster Sum of Squares),...
KMeans n_clusters Selection: Silhouette Score Practice + ...
KMeans n_clusters selection method: calculate silhouette_score and silhouette_samples on candidate cluster numbers (e.g., 2/4/6/8), determine optimal k by...
Python Hand-Written K-Means Clustering on Iris Dataset: F...
Python K-Means clustering implementation: using NumPy broadcasting to compute squared Euclidean distance (distEclud), initializing centroids via uniform...
K-Means Clustering Practice: Self-Implemented Algorithm V...
K-Means clustering provides an engineering workflow that is 'verifiable, reproducible, and debuggable': first use 2D testSet dataset for algorithm verification...
Scikit-Learn Logistic Regression Implementation: max_iter...
When using Logistic Regression in Scikit-Learn, max_iter controls maximum iterations affecting model convergence speed and accuracy. If training doesn't...
K-Means Clustering Guide: From Unsupervised Concepts to I...
K-Means clustering algorithm, comparing supervised vs unsupervised learning (whether labels Y are needed), with engineering applications in customer...
Deep Understanding of Logistic Regression & Gradient Desc...
Logistic Regression (LR) is an important classification algorithm in machine learning, widely used in binary classification tasks like sentiment analysis,...
How to Implement Logistic Regression in Scikit-Learn and ...
As C gradually increases, regularization strength gets smaller, model performance on training and test shows upward trend, until around C=0.8, training...
How to Handle Multicollinearity: Common Problems & Soluti...
When using scikit-learn for linear regression, how to handle multicollinearity in least squares method. Multicollinearity may cause instability in regression...
Ridge Regression and Lasso Regression: Differences, Appli...
Ridge Regression and Lasso Regression are two commonly used linear regression regularization methods for solving overfitting and multicollinearity in machine...
Linear Regression Machine Learning Perspective: Matrix Re...
Linear Regression core chain: unify prediction function y=Xw in matrix form, treat parameter vector w as only unknown; use loss function to characterize...
NumPy Matrix Multiplication Hand-written Multivariate Lin...
pandas DataFrame and NumPy matrix multiplication hand-written multivariate linear regression (linear regression implementation). Core idea is to form normal...
sklearn Decision Tree Pruning Parameters: max_depth/min_s...
Common parameters for decision tree pruning (pre-pruning) in engineering: max_depth, min_samples_leaf, min_samples_split, max_features, min_impurity_decrease...
Confusion Matrix to ROC: Complete Review of Imbalanced Bi...
Confusion matrix (TP, FP, FN, TN) with unified metrics: Accuracy, Precision, Recall (Sensitivity), F1 Measure, ROC curve, AUC value, and practical business interpretation for classification models.
Decision Tree from Split to Pruning: Information Gain/Gai...
Complete chain from 'split' to 'pruning', explain why usually uses greedy algorithm to form 'local optimum', and differences in splitting criteria between...
sklearn Decision Tree Practice: criterion, Graphviz Visua...
Complete flow of DecisionTreeClassifier on load_wine dataset from data splitting, model evaluation to decision tree visualization (2026 version). Focus on...
Decision Tree Model Detailed: Node Structure, Conditional...
Decision Tree model systematic overview for classification tasks: three types of nodes (root/internal/leaf), recursive split flow from root to leaf, and...
Decision Tree Information Gain Detailed: Information Entr...
Decision tree information gain (Information Gain) explained, first using information entropy (Entropy) to explain impurity, then explaining why when splitting...
K-Fold Cross-Validation Practice: sklearn Look at Mean/Va...
Random train/test split causes evaluation metrics to be unstable, and gives engineering solution: K-Fold Cross Validation. Through sklearn's cross_val_score to...
KNN Must Normalize First: Min-Max Correct Method, Data Le...
In scikit-learn machine learning training pipeline, distance-based models like KNN are extremely sensitive to inconsistent feature scales: Euclidean distance...
KNN/K-Nearest Neighbors Algorithm Practice: Euclidean Dis...
KNN/K-Nearest Neighbors Algorithm: From Euclidean distance calculation, distance sorting, TopK voting to function encapsulation, giving reproducible Python...
scikit-learn KNN Practice: KNeighborsClassifier, kneighbo...
From unified API (fit/predict/transform/score) to kneighbors to find K nearest neighbors of test samples, then using learning curve/parameter curve to select...
Data Mining: From Wine Classification to Machine Learning...
2025's most commonly used machine learning concept framework: supervised learning (classification/regression), unsupervised learning (clustering/dimensionality...