{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreifocolltx5hcu3btis7wxvzcovmclwrwcgjhfuuji2coccitp6dee",
"uri": "at://did:plc:4rgrdigiftglskeax4wvmsev/app.bsky.feed.post/3mfbbmch2csv2"
},
"coverImage": {
"$type": "blob",
"ref": {
"$link": "bafkreiflo6xt7is6b2iafwghkjahlgggocme5jwjsbeuqqwcywuvjhmszm"
},
"mimeType": "image/png",
"size": 24783
},
"path": "/abs/2602.17423v1",
"publishedAt": "2026-02-20T01:00:00.000Z",
"site": "https://arxiv.org",
"tags": [
"Afroditi Kolomvaki",
"Fangshuo Liao",
"Evan Dramko",
"Ziyun Guang",
"Anastasios Kyrillidis"
],
"textContent": "**Authors:** Afroditi Kolomvaki, Fangshuo Liao, Evan Dramko, Ziyun Guang, Anastasios Kyrillidis\n\nWe investigate the convergence guarantee of two-layer neural network training with Gaussian randomly masked inputs. This scenario corresponds to Gaussian dropout at the input level, or noisy input training common in sensor networks, privacy-preserving training, and federated learning, where each user may have access to partial or corrupted features. Using a Neural Tangent Kernel (NTK) analysis, we demonstrate that training a two-layer ReLU network with Gaussian randomly masked inputs achieves linear convergence up to an error region proportional to the mask's variance. A key technical contribution is resolving the randomness within the non-linear activation, a problem of independent interest.",
"title": "Convergence Analysis of Two-Layer Neural Networks under Gaussian Input Masking"
}