Machine Learning - Category - Zhiverse

CNN Image Recognition Experiment

Author — Fri, 17 Jul 2026 11:03:26 +0800

The cover image was generated by ChatGPT.

Tip

This project was written in August 2023. The following content primarily documents my initial experience learning and experimenting with deep learning at that time. As such, some explanations and viewpoints have been preserved to reflect my understanding during that period.

The year 2023 marked the rapid rise of generative artificial intelligence. From large language models to a wide range of AI-powered applications, artificial intelligence gradually became part of everyday life, making deep learning one of the most prominent and widely discussed technologies of the year.

For me, 2023 also marked the beginning of my journey into the world of deep learning, as I built and experimented with my first deep learning models.

From Gradient Descent to Stochastic Gradient Descent

Author — Sat, 30 Aug 2025 10:54:32 +0800

Cover image was generated by ChatGPT.

Gradient descent is an iterative method used in most modern models to help find optimal parameters. However, in today’s era of big data, the massive amount of data results in significant time costs for gradient descent. One key goal is to compute parameters quickly without sacrificing accuracy.

Sampling Methods

Author — Tue, 12 Aug 2025 17:17:44 +0800

Cover image generated by ChatGPT.

Why do we need sampling? What are the benefits for data analysis? When building models, we sometimes find that the amount of data required for training is too large, which prolongs the training time. The purpose of sampling is to select a representative finite sample from a large population and perform statistical analysis on it, thereby inferring the characteristics or properties of the overall population.

From this, we can see that a sample is a subset of the population. Sampling methods refer to how we reasonably select a subset of the population as a sample. Since the sample size is smaller than the population, we hope that the sample statistics calculated from the selected sample approximate the population statistics, minimizing the bias introduced by sampling.

Logistic Regression

Author — Wed, 23 Jul 2025 10:55:24 +0800

The cover image was generated by ChatGPT.

Introduction

Logistic regression is a type of binary classification model derived from linear regression. Unlike linear regression, which fits a regression line that passes through the data points by minimizing the sum of squared distances, known as the least squares method. The goal of logistic regression is to find a decision boundary that clearly separates the data into two distinct classes.

Methods for Selecting Models

Author — Sun, 20 Jul 2025 11:55:51 +0800

How can we evaluate and compare multiple models, and select the best one for statistical analysis or machine learning? Simple models may fail to fit the data well, while overly complex models may suffer from "overfitting", reducing their predictive performance on future data.

K-Means Clustering

Author — Fri, 11 Jul 2025 14:49:59 +0800

K-means clustering, often referred to in Chinese as the "k-means algorithm", is a method used to partition $n$ data points into $k$ clusters. Each point is assigned to the cluster whose center (centroid) is closest to it.

Methods of Normalization

Author — Wed, 09 Jul 2025 14:36:01 +0800

Introduction to Perceptron

Author — Tue, 20 May 2025 22:30:24 +0800

The Perceptron is one of the earliest models of artificial neural networks and laid the foundation for modern machine learning, often regarded as the 'ancestor' of neural networks. It performs simple binary classification tasks and mathematically simulates the logic of how neurons operate. Even today, it is frequently used as a teaching example for beginners in machine learning.

Multi-dimensional Scaling

Author — Thu, 15 May 2025 15:58:34 +0800

Multi-dimensional Scaling (MDS) is a dimensionality reduction technique that calculates pairwise distances between objects in a dataset and represents them in a lower-dimensional space while preserving the original relative distance structure. MDS can be seen as a method for visualizing data to intuitively observe the relative relationships between data points.

Definition of Dataset

Author — Fri, 09 May 2025 23:30:30 +0800

In this era of big data, our lives are filled with various forms of information. If we treat each piece of data as an observation and collect them together, we can roughly obtain a dataset composed of many individual data points. These datasets can be either structured or unstructured. By organizing and filtering the data to extract useful information and eliminate irrelevant parts, we can create a dataset suitable for analysis, model training, or academic research. A dataset may include different data types such as numbers, text, images, audio, or video, depending on its intended application and the methods used to collect and organize it.

Methods of Cross-Validation

Author — Wed, 30 Apr 2025 10:42:50 +0800

Cross-validation (CV) is a commonly used model validation technique in machine learning, employed to assess a model’s generalization ability on unseen data. By splitting the dataset into training, validation, and test sets, and repeatedly conducting training and evaluation, cross-validation can effectively test a model’s performance even when the data is limited. It helps evaluate generalization ability and is useful in reducing issues like overfitting and bias. Cross-validation plays a crucial role in controlling overfitting, selecting the best model, and tuning hyperparameters. In particular, it helps predict how a model will perform on new data, making it an indispensable part of modern machine learning workflows.

Introduction to Machine Learning

Author — Wed, 02 Apr 2025 15:16:47 +0800

Machine Learning (ML) is a subfield of Artificial Intelligence (AI) that enables computers to learn from data and make predictions or decisions without being explicitly programmed step by step. In traditional data analysis and programming, developers often need to manually write specific rules for each analytical task. In contrast, machine learning relies on data and uses trained models to automatically identify rules or discover patterns. With the rise of the digital age, massive amounts of information and data are generated and collected every day. Regardless of whether these datasets are inherently valuable or contain subtle, hidden insights, machine learning allows us to uncover and leverage this information. ML technologies are now widely applied across various industries, including finance, healthcare, manufacturing, and autonomous driving.

Principal Component Analysis (PCA)

Author — Tue, 25 Mar 2025 15:09:49 +0800

Principal Component Analysis (PCA) is a commonly used dimension reduction method. It transforms an originally linearly dependent dataset into a set of linearly independent new variables through an orthogonal transformation. This process minimizes the loss of information when projecting the dataset onto these new variables. These new variables are ranked based on their variance, with the one having the highest variance called the first principal component, the second highest called the second principal component, and so on. PCA is widely used to map high-dimensional data into lower-dimensional space while preserving as much of the original data’s key features and information as possible.

Understanding the Confusion Matrix

Author — Fri, 21 Mar 2025 18:04:29 +0800

In machine learning, we can build various models, ranging from simple linear regression to complex neural network models. These models can be applied to different fields, such as image recognition, language generation, and financial forecasting. Each model has its advantages and applicable scenarios. For example, linear regression is suitable for fitting and analyzing datasets with target variables, while deep learning can handle complex nonlinear datasets. However, how do we assess the quality of a model? Simply relying on accuracy may not be sufficient to evaluate model performance, especially in cases with imbalanced class distributions. Therefore, we use **Confusion Matrix** to analyze the performance of classification models.