Alexandr Kalinin

@alxndrkalinin

Biomedical Image Analysis | PhD in Bioinformatics

Ann Arbor, MI
Дата регистрации: апрель 2010 г.

Твиты

Вы внесли @alxndrkalinin в черный список

Вы уверены, что хотите видеть эти твиты? Если вы просто просмотрите твиты, @alxndrkalinin по-прежнему останется в черном списке.

  1. Закрепленный твит
    11 февр.

    Model-Based and Model-Free Techniques for Amyotrophic Lateral Sclerosis Diagnostic Prediction and Patient Clustering code:

    Отменить
  2. ретвитнул(а)
    26 июн.

    I would like to announce that are releasing image augmentation library v0.3.0. To simplify the reproducible we added load / save to python dictionary, json, and yaml config files.

    Отменить
  3. ретвитнул(а)
    25 июн.

    CVPR 2019 Implementations: get all the papers with code in one place here:

    Отменить
  4. ретвитнул(а)
    23 июн.

    All materials of Deep Unsupervised Learning now up: Great semester w/,,, and guest inst. ,,A Efros, Covers: AR / PixelCNN, Flow models, VAE, GAN, self-supervised learning, etc...

    Отменить
  5. ретвитнул(а)
    23 июн.

    resnext101_32x8d_wsl: the ConvNet pre-trained on Instagram hashtags and fine-tuned on ImageNet, yielding a record-breaking 85.4% top-1 accuracy is now available.

    Отменить
  6. ретвитнул(а)
    24 июн.

    We present the adipocyte U-net! Fast and accurate segmentation of adipocytes so we can derive phenotypes from histology imaging data and perform GWAS! We analyse thousands of samples across 4 different cohorts - Including GTEx!

    Показать эту ветку
    Отменить
  7. ретвитнул(а)
    20 июн.

    "Most of the common statistical models (t-test, correlation, ANOVA; chi-square, etc.) are special cases of linear models or a very close approximation. This beautiful simplicity means that there is less to learn."

    Отменить
  8. ретвитнул(а)
    20 июн.

    This morning Anna Kreshuk showed ilastik and teased the upcoming and much anticipated deep-learning additions. Great, engaging, and interactive talk.

    Отменить
  9. ретвитнул(а)
    19 июн.

    The new answer to a 77-year-old problem in data analysis, published today in . Instead of significance tests, use estimation graphics. Our software suite DABEST makes it easy for everyone to visualize effect sizes.

    Показать эту ветку
    Отменить
  10. ретвитнул(а)
    16 июн.

    [Public LB] 6th place solution of Freesound Audio Tagging 2019 competition Key points: * CNN model with attention, skip connections and auxiliary classifiers * SpecAugment, Mixup augmentations * Ensemble with MLP second-level model

    Отменить
  11. ретвитнул(а)
    15 июн.

    ok we are writing a lab handbook so I spent my afternoon reading 25 lab handbooks to figure out all the things that people put in lab handbooks. here they are so that you don't have to read 25 lab handbooks. (this is not exhaustive and many of the topics overlap, obvi) 1/

    Показать эту ветку
    Отменить
  12. ретвитнул(а)
    14 июн.

    Really excited to release Bayesian Deep Learning Benchmarks - please share with others who you think might like this, and have a look at the blog/repo/colab: This work was done over a period of a year and a half by many collaborators

    Отменить
  13. ретвитнул(а)
    14 июн.

    Check out Evolved Transformer, a new state-of-the-art Transformer architecture for natural language processing, derived using a fresh approach to evolution-based neural architecture search. Learn all about it here →

    Отменить
  14. 14 июн.

    Gene- and Pathway-Based Deep Neural Network for Multi-omics Data Integration to Predict Cancer Survival Outcomes code:

    Отменить
  15. 13 июн.

    2nd and 3rd place solutions also used Mask-RCNN and a Hybrid Task Cascade with different scheduling, TTA, post-processing, and ensembling 2nd place: 3rd place:

    Показать эту ветку
    Отменить
  16. 13 июн.

    1st place solution for iMaterialist (Fashion) 2019 @ FGVC6 code: - Hybrid Task Cascade w/ ResNeXt-101-64x4d-FPN backbone - pre-train from SOTA on COCO - customized TTA & ensembling - simple post-processing

    Показать эту ветку
    Отменить
  17. 13 июн.

    Voilà: from notebooks to standalone applications & dashboards - supports Jupyter interactive widgets - doesn't permit arbitrary code execution - language-agnostic, built upon Jupyter std protocols & file formats - extensible w/ a template system

    Отменить
  18. 13 июн.

    1st place solution for 's Google Landmark Retrieval 2019 - data auto-cleaning w/ train labels, global descriptors, spatial verification - FishNet-150, ResNet-101, SEResNeXt-101 w/ ArcFace & CosFace losses - top-k similarity post-processing

    Отменить
  19. ретвитнул(а)
    12 июн.

    Weight Agnostic Neural Networks 🦎 Inspired by precocial species in biology, we set out to search for neural net architectures that can already (sort of) perform various tasks even when they use random weight values. Article: PDF:

    Показать эту ветку
    Отменить
  20. ретвитнул(а)
    7 июн.

    In order to feel like an artist you need: 40k paintings from wikiart, 2080ti x 2, 4 days of StyleGAN training

    Показать эту ветку
    Отменить
  21. ретвитнул(а)
    11 июн.

    Thanks to for this list of recommended medical vocabularies: SNOMED CT for clinical diagnoses; ICD9/10 for billing diagnoses; RxNorm for drugs; CPT4 and HCPCS for procedures; and LOINC for lab tests

    Показать эту ветку
    Отменить

Загрузка может занять некоторое время.

Вероятно, серверы Твиттера перегружены или в их работе произошел кратковременный сбой. Повторите попытку или посетите страницу Статус Твиттера, чтобы узнать более подробную информацию.

    Вам также может понравиться

    ·