Yang

Introduction to Classic Deepfake Detection Models

Fri, 06 Mar 2026 21:35:00 +0800

Introduction

With the rapid development of Generative Adversarial Networks (GANs) and Diffusion Models, Deepfake content is becoming increasingly prevalent on the internet. Effectively identifying these forged images and videos has become a critical issue in multimedia forensics and information security.

Core Content

This article introduces some milestone works in the field of Deepfake detection:

1. Spatial Feature-based Models

MesoNet: Focuses on macroscopic burial artifacts in compressed facial images.
Xception-based (FaceForensics++): A representative of transfer learning, fine-tuning models pre-trained on large-scale datasets.

2. Temporal Feature-based Models

Deepfake Stacked RNN: Utilizes the continuity between video frames to capture forgery traces.

3. Frequency Domain-based Models

F3-Net: Identifies forgeries through frequency domain decomposition and frequency statistics.

4. Biological Feature-based Models

Lip-sync Check: Observing whether lip movements are synchronized with speech.
Blink Detection: Early Deepfake models often struggled to generate natural blinking.

Reflections

References

Attention Is All You Need — Deep Dive into the Transformer Architecture

Fri, 06 Mar 2026 20:00:00 +0800

📄 Background

Before Transformers, sequence-to-sequence tasks relied heavily on RNN/LSTM architectures, which suffered from two major bottlenecks:

Sequential computation prevents parallelization
Long-range dependency degradation over long sequences

Vaswani et al. proposed the Transformer at NeurIPS 2017, relying entirely on attention mechanisms to model global dependencies — no recurrence, no convolution.

🔑 Core Mechanisms

Self-Attention

For input $X \in \mathbb{R}^{n \times d}$, we compute Query, Key, Value projections:

$$Q = XW^Q, \quad K = XW^K, \quad V = XW^V$$

Research Insights: How to Read Papers Efficiently

Thu, 05 Mar 2026 18:00:00 +0800

The Three-Pass Approach

A classic method from Keshav’s “How to Read a Paper”:

Pass 1: 5–10 minutes — Get the big picture

Read: title, abstract, introduction
Scan all figure captions and conclusions
Decide: is this worth reading in depth?

Pass 2: ~1 hour — Grasp the content

Read the body, skip proofs
Focus on experimental setup and key figures
Note limitations the authors acknowledge

Pass 3: Deep read — Replication-level understanding

Derive every equation
Challenge every assumption

My Note Template

## Problem
## Key Insight
## Method
## Experiments
## My Take
## Action Items

Recommended Tools

Tool	Purpose
Zotero	Reference management
Obsidian	Connected note-taking
Connected Papers	Citation graph visualization
Semantic Scholar	AI-assisted literature search

Book Recommendation: Designing Data-Intensive Applications (DDIA)

Mon, 23 Feb 2026 11:43:00 +0800

📚 About the Book

“Designing Data-Intensive Applications” (often referred to as DDIA) by Martin Kleppmann is widely considered a masterpiece in the field of distributed systems and data engineering.

Modern applications are increasingly “data-intensive” rather than merely “compute-intensive”. This book systematically explores the core technologies behind storing and processing data, covering relational databases, NoSQL, stream processing, and batch processing. Instead of focusing on specific tech stacks or vendors, the author breaks down complex backend mechanisms by focusing on three overarching goals that any robust architecture must meet: Reliability, Scalability, and Maintainability. It offers an incredibly lucid breakdown of core distributed concepts like replication, partitioning, transactions, and consensus correctness.

Remote Private Networking with Tailscale

Sat, 21 Feb 2026 18:59:24 +0800

Tailscale Remote Network

Introduction

Sometimes we build tools without a full authentication system, or we need to access private network resources from another location. In this case, Tailscale is a simple way to build a secure virtual LAN across different networks.

What is Tailscale?

Tailscale is a free private networking solution built on WireGuard. It provides an easy-to-use experience for remote access and private mesh networking.

Install Tailscale

Linux is used as the example here:

Personal Agent Assistant - Openclaw Installation Guide

Wed, 11 Feb 2026 20:14:06 +0800

Install Openclaw

Create an `openclaw` user to limit permissions

Create a dedicated openclaw user with sudo privileges:

sudo adduser --disabled-password --gecos "" openclaw
sudo usermod -aG sudo openclaw # Optional: keep sudo access for this user
sudo passwd openclaw # Set password, needed later

Install with the official script

You will be asked to enter the password during installation:

curl -fsSL https://openclaw.ai/install.sh | bash

Complete installation in the setup wizard

Follow the wizard based on your needs.

Quick Setup - Docker Installation

Wed, 11 Feb 2026 10:43:31 +0800

Install Docker on Ubuntu

1. Update the server

sudo apt-get update
sudo apt-get upgrade

2. Install dependencies and add the official repository

sudo apt update
sudo apt install -y ca-certificates curl gnupg lsb-release

# Add Docker official GPG key
sudo mkdir -p /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg

# Add Docker official repository
echo \
 "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] \
 https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" \
 | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

3. Reboot the server to update the kernel

sudo reboot

4. Install Docker

sudo apt update
sudo apt install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

5. Verify installation

sudo docker run hello-world
sudo docker rm $(sudo docker ps -a -q --filter ancestor=hello-world)
sudo docker rmi hello-world

6. Configure a registry mirror

sudo mkdir -p /etc/docker
sudo tee /etc/docker/daemon.json <<EOF
{
 "registry-mirrors": ["https://docker.m.daocloud.io"]
}
EOF
sudo systemctl daemon-reload
sudo systemctl restart docker

Common commands

# Start
docker compose up -d --build
# Restart
docker compose restart
# Stop
docker compose down
# View logs
docker compose logs -f
# List running containers
docker ps
# List all containers
docker ps -a
# Remove a container
docker rm <container_id>
# Remove all containers
docker rm $(docker ps -a -q)
# Remove an image
docker rmi <image_id>
# Remove all images
docker rmi $(docker images -q)
# Remove all containers and images
docker system prune -a

7. Common issues

Cannot access the official repository

You can try switching to the Aliyun mirror:

Yang

Introduction to Classic Deepfake Detection Models

Introduction

Core Content

1. Spatial Feature-based Models

2. Temporal Feature-based Models

3. Frequency Domain-based Models

4. Biological Feature-based Models

Reflections

References

Attention Is All You Need — Deep Dive into the Transformer Architecture

📄 Background

🔑 Core Mechanisms

Self-Attention

Research Insights: How to Read Papers Efficiently

The Three-Pass Approach

Pass 1: 5–10 minutes — Get the big picture

Pass 2: ~1 hour — Grasp the content

Pass 3: Deep read — Replication-level understanding

My Note Template

Recommended Tools

Book Recommendation: Designing Data-Intensive Applications (DDIA)

📚 About the Book

Remote Private Networking with Tailscale

Tailscale Remote Network

Introduction

What is Tailscale?

Install Tailscale

Personal Agent Assistant - Openclaw Installation Guide

Install Openclaw

Create an openclaw user to limit permissions

Install with the official script

Complete installation in the setup wizard

Quick Setup - Docker Installation

Install Docker on Ubuntu

1. Update the server

2. Install dependencies and add the official repository

3. Reboot the server to update the kernel

4. Install Docker

5. Verify installation

6. Configure a registry mirror

Common commands

7. Common issues

Cannot access the official repository

Create an `openclaw` user to limit permissions