Enabling Fast Differentially Private SGD via Static Graph Compilation and Batch-Level Parallelism

by

· Dec 6, 2020 · 6 views ·

NeurIPS

<div id="presentation-embed-38940933"></div>
<script src='https://slideslive.com/embed_presentation.js'></script>
<script>
    embed = new SlidesLiveEmbed('presentation-embed-38940933', {
        presentationId: '38940933',
        autoPlay: false, // change to true to autoplay the embedded presentation
        verticalEnabled: true
    });
</script>