wubang
User Content
请将下图中所有英文文字替换为对应的中文，图片的布局、结构、颜色、箭头、方框样式完全不变，仅替换文字内容。具体替换对应关系如下： - "True label" → "真实标签" - "Local data" → "本地数据" - "Pred" → "预测" - "Loss" → "损失" - "Trained with E local epochs" → "本地训练 E 个轮次" - "Upload local model parameters" → "上传本地模型参数" - "Client $\mathcal{C}_m$" → "客户端 $\mathcal{C}_m$" - "Local model parameters $W_m = W$" → "本地模型参数 $W_m = W$" - "$\mathcal{C}_m$ Download global model" → "$\mathcal{C}_m$ 下载全局模型" - "Initialize global model" → "初始化全局模型" - "Aggregate the global model" → "聚合全局模型" - "Server $\mathcal{S}$" → "服务端 $\mathcal{S}$" - "Global model parameters $W$" → "全局模型参数 $W$" - "Attacker download $\mathcal{C}_m$'s local model parameters" → "攻击者下载 $\mathcal{C}_m$ 的本地模型参数" - "Dummy label" → "虚假标签" - "Dummy data" → "虚假数据" - "Attacker" → "攻击者" - "Optimize" → "优化" - "Gradient matching" → "梯度匹配" - "Fig. 1. Gradients inversion attacks in federated learning." → "图1 联邦学习中的梯度反转攻击"
请将下图中所有英文文字替换为对应的中文，图片的布局、结构、颜色、箭头、方框样式完全不变，仅替换文字内容。具体替换对应关系如下： - "True label" → "真实标签" - "Local data" → "本地数据" - "Pred" → "预测" - "Loss" → "损失" - "Trained with E local epochs" → "本地训练 E 个轮次" - "Upload local model parameters" → "上传本地模型参数" - "Client $\mathcal{C}_m$" → "客户端 $\mathcal{C}_m$" - "Local model parameters $W_m = W$" → "本地模型参数 $W_m = W$" - "$\mathcal{C}_m$ Download global model" → "$\mathcal{C}_m$ 下载全局模型" - "Initialize global model" → "初始化全局模型" - "Aggregate the global model" → "聚合全局模型" - "Server $\mathcal{S}$" → "服务端 $\mathcal{S}$" - "Global model parameters $W$" → "全局模型参数 $W$" - "Attacker download $\mathcal{C}_m$'s local model parameters" → "攻击者下载 $\mathcal{C}_m$ 的本地模型参数" - "Dummy label" → "虚假标签" - "Dummy data" → "虚假数据" - "Attacker" → "攻击者" - "Optimize" → "优化" - "Gradient matching" → "梯度匹配" - "Fig. 1. Gradients inversion attacks in federated learning." → "图1 联邦学习中的梯度反转攻击"
请将下图中所有英文文字替换为对应的中文，图片的布局、结构、颜色、箭头、方框样式完全不变，仅替换文字内容。具体替换对应关系如下： - "True label" → "真实标签" - "Local data" → "本地数据" - "Pred" → "预测" - "Loss" → "损失" - "Trained with E local epochs" → "本地训练 E 个轮次" - "Upload local model parameters" → "上传本地模型参数" - "Client $\mathcal{C}_m$" → "客户端 $\mathcal{C}_m$" - "Local model parameters $W_m = W$" → "本地模型参数 $W_m = W$" - "$\mathcal{C}_m$ Download global model" → "$\mathcal{C}_m$ 下载全局模型" - "Initialize global model" → "初始化全局模型" - "Aggregate the global model" → "聚合全局模型" - "Server $\mathcal{S}$" → "服务端 $\mathcal{S}$" - "Global model parameters $W$" → "全局模型参数 $W$" - "Attacker download $\mathcal{C}_m$'s local model parameters" → "攻击者下载 $\mathcal{C}_m$ 的本地模型参数" - "Dummy label" → "虚假标签" - "Dummy data" → "虚假数据" - "Attacker" → "攻击者" - "Optimize" → "优化" - "Gradient matching" → "梯度匹配" - "Fig. 1. Gradients inversion attacks in federated learning." → "图1 联邦学习中的梯度反转攻击"
请将下图中所有英文文字替换为对应的中文，图片的布局、结构、颜色、箭头、方框样式完全不变，仅替换文字内容。具体替换对应关系如下： - "True label" → "真实标签" - "Local data" → "本地数据" - "Pred" → "预测" - "Loss" → "损失" - "Trained with E local epochs" → "本地训练 E 个轮次" - "Upload local model parameters" → "上传本地模型参数" - "Client $\mathcal{C}_m$" → "客户端 $\mathcal{C}_m$" - "Local model parameters $W_m = W$" → "本地模型参数 $W_m = W$" - "$\mathcal{C}_m$ Download global model" → "$\mathcal{C}_m$ 下载全局模型" - "Initialize global model" → "初始化全局模型" - "Aggregate the global model" → "聚合全局模型" - "Server $\mathcal{S}$" → "服务端 $\mathcal{S}$" - "Global model parameters $W$" → "全局模型参数 $W$" - "Attacker download $\mathcal{C}_m$'s local model parameters" → "攻击者下载 $\mathcal{C}_m$ 的本地模型参数" - "Dummy label" → "虚假标签" - "Dummy data" → "虚假数据" - "Attacker" → "攻击者" - "Optimize" → "优化" - "Gradient matching" → "梯度匹配" - "Fig. 1. Gradients inversion attacks in federated learning." → "图1 联邦学习中的梯度反转攻击"