{
    "146703-21899508":
    {
        "datetime_modified":"20260403T100524",
        "datetime_start":"20260417T100000",
        "datetime_end":"20260417T110000",
        "has_end_time":1,
        "date_start":"2026-04-17",
        "date_end":"2026-04-17",
        "time_start":"10:00:00",
        "time_end":"11:00:00",
        "time_zone":"America\/Detroit",
        "event_title":"Statistics Department Seminar Series: Yuting Wei, Associate Professor, Department of Statistics & Data Science, University of Pennsylvania",
        "occurrence_title":"",
        "combined_title":"Statistics Department Seminar Series: Yuting Wei, Associate Professor, Department of Statistics & Data Science, University of Pennsylvania: Efficient Sampling with Diffusion Models: Sharp and Adaptive Guarantees",
        "event_subtitle":"Efficient Sampling with Diffusion Models: Sharp and Adaptive Guarantees",
        "event_type":"Workshop \/ Seminar",
        "event_type_id":"21",
        "description":"The score-based diffusion models have become a cornerstone of modern generative AI. While recent works aim to develop sharp convergence guarantees, the iteration complexity of existing analyses typically scales with the ambient data dimension $d$ of the target distribution, leading to overly conservative theory that fails to explain its practical efficiency. This motivates us to understand how diffusion models can achieve sampling speed-ups through automatic exploitation of intrinsic low dimensionality of data for both continuous and discrete distributions.  \n\nThis talk explores two key scenarios: (1) For a broad class of continous distributions with intrinsic dimension $k$, we show that the iteration complexity of the denoising diffusion probabilistic model (DDPM) scales nearly linearly with $k$, which is optimal under the KL divergence metric; (2) For masking discrete diffusions, under a continuous-time Markov chain (CTMC) formulation, we introduce a modified $\\tau$-leaping sampler whose convergence rate is governed by an intrinsic information-theoretic quantity, termed the \\emph{effective total correlation}, which is upper bounded by $d \\log S$ (with $S$ the vocabulary size) but can be sublinear or even constant for structured discrete distributions.",
        "occurrence_notes":null,
                "guid":"146703-21899508@events.umich.edu",
        "permalink":"http:\/\/events.umich.edu\/event\/146703",
        "building_id":"1000167",
        "building_name":"West Hall",
        "campus_maps_id":"163",
        "room":"340",
        "location_name":"West Hall",
        "has_livestream":0,
        "cost":"",
        "tags":["seminar"],
        "website":"https:\/\/lsa.umich.edu\/stats",
        "sponsors":[
             {
                "group_name":"Department of Statistics Seminar Series",
                "group_id":"3585",
                "website":""                }                    ],
        "image_url":"https:\/\/events.umich.edu\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
        "styled_images":{
                                        "event_thumb":"https:\/\/events.umich.edu\/media\/cache\/event_thumb\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_large":"https:\/\/events.umich.edu\/media\/cache\/event_large\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_large_2x":"https:\/\/events.umich.edu\/media\/cache\/event_large_2x\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_large_lightbox":"https:\/\/events.umich.edu\/media\/cache\/event_large_lightbox\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "group_thumb":"https:\/\/events.umich.edu\/media\/cache\/resolve\/group_thumb\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "group_thumb_square":"https:\/\/events.umich.edu\/media\/cache\/resolve\/group_thumb_square\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "group_large":"https:\/\/events.umich.edu\/media\/cache\/resolve\/group_large\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "group_large_lightbox":"https:\/\/events.umich.edu\/media\/cache\/resolve\/group_large_lightbox\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_large_crop":"https:\/\/events.umich.edu\/media\/cache\/resolve\/event_large_crop\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_list":"https:\/\/events.umich.edu\/media\/cache\/resolve\/event_list\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_list_2x":"https:\/\/events.umich.edu\/media\/cache\/event_list_2x\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_grid":"https:\/\/events.umich.edu\/media\/cache\/event_grid\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_grid_2x":"https:\/\/events.umich.edu\/media\/cache\/resolve\/event_grid_2x\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_feature_large":"https:\/\/events.umich.edu\/media\/cache\/resolve\/event_feature_large\/media\/attachments\/2026\/04\/event_146703_original-1.jpg",
                                            "event_feature_thumb":"https:\/\/events.umich.edu\/media\/cache\/resolve\/event_feature_thumb\/media\/attachments\/2026\/04\/event_146703_original-1.jpg"                    },
        "occurrence_count":1,
        "first_occurrence":21899508
    }    }