https://shio-chan-dev.github.io/jeanblog/zh/categories/representation-learning/Recent content in Representation Learning on Jeanphilo BlogHugo -- 0.159.2zh-cnSat, 24 Jan 2026 13:22:02 +0800https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/1-contrastive-loss-function/Sat, 24 Jan 2026 13:22:02 +0800https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/1-contrastive-loss-function/从公式到实验，系统理解对比损失（Contrastive Loss）如何拉近正样本、推远负样本。https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/2-triplet-loss-function/Sat, 24 Jan 2026 13:22:02 +0800https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/2-triplet-loss-function/从 anchor-positive-negative 视角理解 Triplet Loss，并用最小可运行实验掌握 hard negative mining。https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/3-infonce-simclr/Sat, 24 Jan 2026 13:22:02 +0800https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/3-infonce-simclr/从 InfoNCE 公式到 SimCLR 训练流程，理解自监督对比学习的关键设计。https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/4-clip-contrastive-learning-objective/Sat, 24 Jan 2026 13:22:02 +0800https://shio-chan-dev.github.io/jeanblog/zh/ai/contrastive-learning/4-clip-contrastive-learning-objective/从损失函数视角理解 CLIP 的双向对比学习目标，建立跨模态对齐的核心直觉。