Deep Learning Academy · 课时

处理不平衡数据的类别权重

防止多数类占据主导

处理不平衡数据的类别权重 是 CoddyKit 上的免费 Deep Learning Academy 课时。这是第 4 节课，共 4 节。你可以在下方免费阅读本课时的完整内容 — 然后在浏览器中使用内置代码编辑器和全天候 AI 导师进行实践。这是 Deep Learning Academy 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 Deep Learning Academy 课程共包含 4 节课。

本课时的部分内容尚未翻译，以英文显示。

When One Class Dominates

If 95 percent of your samples are one class, the model can score high by always guessing it. That is the class imbalance trap. ⚖️

Why Accuracy Lies Here

A lazy model that ignores the rare class still looks 95 percent accurate. So accuracy hides total failure on the minority you actually care about.

The Loss Follows the Majority

Each sample contributes equally to the loss, so the abundant class pulls the gradient hardest. The model learns mostly from the majority class.

Reweight the Loss

The fix is to give rare classes a heavier voice. Class weights multiply each sample's loss so minority mistakes count for more.

Bigger Weight for Rarer Class

A common recipe sets each weight inversely proportional to its class frequency. The rarer the class, the larger its weight grows.

Pass Weights to CrossEntropy

For multiclass tasks, hand a weight tensor of length C to the loss. Each entry scales the penalty for that class's errors.

import torch.nn as nn
w = torch.tensor([1.0, 9.0])
loss_fn = nn.CrossEntropyLoss(weight=w)

pos_weight for Binary Tasks

For two-class problems, BCEWithLogitsLoss takes pos_weight, a single number that boosts the positive class to offset its rarity.

loss_fn = nn.BCEWithLogitsLoss(pos_weight=torch.tensor([9.0]))

Sample, Don't Just Weight

Weighting is not the only tool. You can also oversample the rare class or undersample the common one to balance each batch.

Balance Batches With a Sampler

A WeightedRandomSampler draws minority samples more often, so every batch the DataLoader serves is roughly balanced by design.

from torch.utils.data import WeightedRandomSampler

Focal Loss for Hard Cases

For extreme imbalance, focal loss down-weights easy, well-classified examples so the model focuses on the hard, rare ones.

Judge by the Right Metric

After rebalancing, drop accuracy and track F1 or precision and recall. They reveal whether the rare class is finally being learned.

Quick Check

Your dataset is 95 percent negative. How do you stop the model from ignoring positives?

Recap: Give the Rare Class a Voice

Imbalance lets the majority hijack training, so you fight back with class weights, balanced sampling, or focal loss, and you judge results with F1. 💪

常见问题解答

「处理不平衡数据的类别权重」课时是免费的吗？

是的 — 「处理不平衡数据的类别权重」的完整文本可在网页上免费阅读。要进行交互式练习（内置代码编辑器和全天候 AI 导师）并解锁 Deep Learning Academy 课程的其余内容，请升级到 CoddyKit PRO。 Deep Learning Academy 课程共包含 4 节课。

「处理不平衡数据的类别权重」这节课中我会学到什么？

防止多数类占据主导你通过在浏览器中直接运行的动手代码来练习 Deep Learning Academy，全天候 AI 导师会在你学习这节课的过程中回答你的问题。

学习 Deep Learning Academy 需要有经验吗？

无需任何先前经验。CoddyKit 上的 Deep Learning Academy 课程适合初学者到高级学习者，你可以从这里开始或从头开始，按照自己的节奏学习。这是第 4 节课，共 4 节。

「处理不平衡数据的类别权重」课时需要多长时间？

大多数 CoddyKit 课程大约需要 5–10 分钟。每节课都很精短且互动，所以你能稳步进步，并在网页和应用中从离开的地方继续。

我能在这节 Deep Learning Academy 课中编写并运行代码吗？

能。每节 Deep Learning Academy 课都包含内置代码编辑器，你可以在浏览器中直接编写并运行真实代码，并获得即时 AI 反馈 — 无需本地设置。

此课程中的所有课时

← 返回 Deep Learning Academy