- 博客(10)
- 收藏
- 关注
RSS订阅原创 Flow Matching学习笔记
假设存在目标分布的样本x1x_1x1，构造一个依赖于时间的条件分布pt(x∣x1)p_t(x|x_1)pt(x∣x1)满足考虑更一般的条件分布形式，假设有一个隐变量zzz，条件分布的Flow Matching为LCFM(θ)=Et∼U(0,1),x∼pt(x∣z),z∼(qz)∣∣vθ(t,x)−ut(x∣z)∣∣2L_{CFM}(\theta)=E_{t\sim U(0,1),x\sim p_t(x|z),z\sim (q_z)}||v_\theta(t,x)-u_t(x|z)||^2LCFM(
2024-11-26 12:10:24
1992
原创 ICML2023论文阅读记录 - Straightening Out the Straight-Through Estimator
ICML2023论文阅读记录 - Straightening Out the Straight-Through Estimator: Overcoming Optimization Challenges in Vector Quantized Networks
2024-05-14 10:16:56
1235
1
原创 ICCV2023论文阅读记录 - Online Clustered Codebook
ICCV2023论文阅读记录 - Online Clustered Codebook
2024-04-23 10:52:09
2363
1
原创 Interspeech20022论文阅读记录1 - Cross-Layer Similarity Knowledge Distillation for Speech Enhancement
Interspeech2022的论文 Cross-Layer Similarity Knowledge Distillation for Speech Enhancement 的阅读记录。
2022-12-09 12:04:38
674
1
原创 ICASSP2020论文阅读记录1 - 基于VAE的鲁棒视听语音增强(audio-visual speech enhancement)方法
ICASSP2020 视听语音增强文章 Robust unsupervised audio-visual speech enhancement using a mixture of variational autoencoders论文阅读记录
2022-11-30 10:24:33
892
1
原创 ICASSP2022论文阅读记录3 - TalkingFlow
ICASSP2022论文阅读记录：TalkingFlow: Talking Facial Landmark generation with multi-scale normalizing flow network
2022-07-04 11:42:46
1195
1
原创 ICASPP2022论文阅读记录2 - Transformer-S2A
ICASSP2022论文Transformer-S2A的论文阅读记录。
2022-06-29 17:56:51
1525
1
原创 ICASPP2022论文阅读记录1 - VCVTS
ICASSP2022论文阅读记录: VCVTS: Multi-speaker video-to-speech synthesis via cross-modal knowledge transfer from voice conversion
2022-06-23 11:30:25
1437
1
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人