The expansion of AI involves several hot issues related to the workforce (data supply demands, disappearing professions, and undesirable working conditions) that can be overcome with the right crowdsourcing methodology. Nowadays it is important to focus on modern and effective ways to overcome these issues and keep working on critical aspects of successful data collection and labeling.
Yandex Toloka team will present “Crowd Science Workshop: Remoteness, Fairness, and Mechanisms as Challenges of Data Supply by Humans for Automation” at the world’s number one conference on machine learning — NeurIPS 2020. Speakers from all over the world will discuss the future of crowdsourcing markets as well as some topics that were never brought up before:
- Remoteness. A discussion about effectiveness and efficiency of remote work on crowdsourcing platforms.
- Fairness. How the working environment (e.g., a crowdsourcing platform) may help provide executors flexibility in choosing/switching tasks and working hours.
- Mechanisms. Discussion on bilateral mechanisms that not only provide flexibility to the performers, but also guarantee the quality of the result and the efficiency of the process to the customers.
Click the link for more info on Toloka’s workshop at NeurIPS 2020.
Yandex Toloka team will present “Crowd Science Workshop: Remoteness, Fairness, and Mechanisms as Challenges of Data Supply by Humans for Automation” at the world’s number one conference on machine learning — NeurIPS 2020. Speakers from all over the world will discuss the future of crowdsourcing markets as well as some topics that were never brought up before:
- Remoteness. A discussion about effectiveness and efficiency of remote work on crowdsourcing platforms.
- Fairness. How the working environment (e.g., a crowdsourcing platform) may help provide executors flexibility in choosing/switching tasks and working hours.
- Mechanisms. Discussion on bilateral mechanisms that not only provide flexibility to the performers, but also guarantee the quality of the result and the efficiency of the process to the customers.
Click the link for more info on Toloka’s workshop at NeurIPS 2020.
Toloka: Data solutions to drive AI
Crowd Science Workshop at NeurIPS 2020
Remoteness, fairness, and mechanisms as challenges of data supply by humans for automation.
👍1
New scandal around ethic AI
https://www.technologyreview.com/2020/12/04/1013294/google-ai-ethics-research-paper-forced-out-timnit-gebru/
https://www.technologyreview.com/2020/12/04/1013294/google-ai-ethics-research-paper-forced-out-timnit-gebru/
MIT Technology Review
We read the paper that forced Timnit Gebru out of Google. Here’s what it says.
The company's star ethics researcher highlighted the risks of large language models, which are key to Google's business.
Rising ticket prices tomorrow! Have time to register for the best price 🙌
On December 15, Enrico Santus, PhD will hold the webinar on “NLP in Healthcare: Challenges and Opportunities”. He will describe recent research about how Natural Language Processing — the linguistic branch of AI — can improve the healthcare system, increasing its efficiency and efficacy.
💥 Report language: English
Registration — https://lnkd.in/dBKHnxj
On December 15, Enrico Santus, PhD will hold the webinar on “NLP in Healthcare: Challenges and Opportunities”. He will describe recent research about how Natural Language Processing — the linguistic branch of AI — can improve the healthcare system, increasing its efficiency and efficacy.
💥 Report language: English
Registration — https://lnkd.in/dBKHnxj
Data Science UA
NLP in Healthcare: Challenges and Opportunities - Data Science UA
Fastspeech2 - Fast Text to Speech
https://fastspeech2.github.io/fastspeech2/
📄 https://arxiv.org/abs/2006.04558
📦 https://github.com/ming024/FastSpeech2
https://fastspeech2.github.io/fastspeech2/
📄 https://arxiv.org/abs/2006.04558
📦 https://github.com/ming024/FastSpeech2
GitHub
GitHub - ming024/FastSpeech2: An implementation of Microsoft's "FastSpeech 2: Fast and High-Quality End-to-End Text to Speech"
An implementation of Microsoft's "FastSpeech 2: Fast and High-Quality End-to-End Text to Speech" - ming024/FastSpeech2
Fresh lib for PyTorch from Facebook that can help you with large scale parallel trainings
https://github.com/facebookresearch/fairscale
https://github.com/facebookresearch/fairscale
GitHub
GitHub - facebookresearch/fairscale: PyTorch extensions for high performance and large scale training.
PyTorch extensions for high performance and large scale training. - facebookresearch/fairscale
Intro to Machine Learning Course (Good Math needed)
https://www.epfl.ch/labs/mlo/machine-learning-cs-433/
https://www.epfl.ch/labs/mlo/machine-learning-cs-433/
EPFL
Machine Learning CS-433
This course is offered jointly by the TML and MLO groups. Previous year’s website: ML 2023. See here for the ML4Science projects. Contact us: Use the discussion forum. You can also email the head assistant Corentin Dumery, and CC both instructors. Instructors:…
Как научить компьютер понимать человека? А наладить бизнес-процессы с помощью машинного обучения? Что нужно для создания голосового помощника? А чат-бота как сделать?
Такие задачи решают на Факультете обработки естественного языка (NLP-разработки) GeekBrains. Студенты с нуля изучают алгоритмы машинного обучения и основы программирования. Анализируют запросы, учат нейросеть работать с текстом, пишут свои программы. И после выпуска устраиваются на работу в топовые компании.
Что изучают на факультете
• Математические методы → анализ и моделирование.
• Работу с текстами → машинный перевод.
• Программирование на Python, библиотеки и фреймворки.
• Архитектуры MVP-решений.
• Создание чат-ботов.
О том, как получить востребованную профессию → по ссылке
Такие задачи решают на Факультете обработки естественного языка (NLP-разработки) GeekBrains. Студенты с нуля изучают алгоритмы машинного обучения и основы программирования. Анализируют запросы, учат нейросеть работать с текстом, пишут свои программы. И после выпуска устраиваются на работу в топовые компании.
Что изучают на факультете
• Математические методы → анализ и моделирование.
• Работу с текстами → машинный перевод.
• Программирование на Python, библиотеки и фреймворки.
• Архитектуры MVP-решений.
• Создание чат-ботов.
О том, как получить востребованную профессию → по ссылке
This media is not supported in your browser
VIEW IN TELEGRAM
This media is not supported in your browser
VIEW IN TELEGRAM
Sketch Generation with Drawing Process Guided by Vector Flow and Grayscale
📦 https://github.com/TZYSJTU/Sketch-Generation-with-Drawing-Process-Guided-by-Vector-Flow-and-Grayscale
📃 https://arxiv.org/pdf/2012.09004.pdf
📦 https://github.com/TZYSJTU/Sketch-Generation-with-Drawing-Process-Guided-by-Vector-Flow-and-Grayscale
📃 https://arxiv.org/pdf/2012.09004.pdf
ffhq_samples.png
2.5 MB
Soft-IntroVAE: Analyzing and Improving Introspective Variational Autoencoders
https://taldatech.github.io/soft-intro-vae-web/
📦 https://github.com/taldatech/soft-intro-vae-pytorch
📃 https://arxiv.org/abs/2012.13253v1
https://taldatech.github.io/soft-intro-vae-web/
📦 https://github.com/taldatech/soft-intro-vae-pytorch
📃 https://arxiv.org/abs/2012.13253v1
Happy New Year!
Hope it will bring you new ideas, interesting projects, more cool papers, and bring less pain in the ass than previous)
And dont forget to help animals that live on the streets or suffer, they need you.
Hope it will bring you new ideas, interesting projects, more cool papers, and bring less pain in the ass than previous)
And dont forget to help animals that live on the streets or suffer, they need you.
Дата-инженер — ключевой игрок в любой команде, где есть аналитика. Этот специалист отвечает за сбор, хранение и удобную работу с данными. То есть помогает компании решать задачи быстрее и эффективнее. Спрос на таких айтишников очень высокий. Как и уровень их заработной платы.
Хотите стать крутым и незаменимым? Приходите на факультет Data Engineering от GeekBrains.
Чему научат:
— Автоматизировать технические процессы.
— Создавать конвейеры обработки данных.
— Разрабатывать архитектуру хранения данных.
— Настраивать мониторинги.
— Готовить данные для аналитиков.
Будет много практики от опытных дата-инженеров. Отработаете реальные задачи, соберете готовое портфолио, а эйчары из GeekBrains помогут вам найти работу!
Регистрируйтесь по ссылке → https://geekbrains.ru/link/v34bmh
Хотите стать крутым и незаменимым? Приходите на факультет Data Engineering от GeekBrains.
Чему научат:
— Автоматизировать технические процессы.
— Создавать конвейеры обработки данных.
— Разрабатывать архитектуру хранения данных.
— Настраивать мониторинги.
— Готовить данные для аналитиков.
Будет много практики от опытных дата-инженеров. Отработаете реальные задачи, соберете готовое портфолио, а эйчары из GeekBrains помогут вам найти работу!
Регистрируйтесь по ссылке → https://geekbrains.ru/link/v34bmh
LambdaNetworks: Modeling long-range Interactions without Attention
Got SOTA on ImageNet
📄 https://openreview.net/forum?id=xTJEN-ggl1b
📦 https://github.com/leaderj1001/LambdaNetworks
Got SOTA on ImageNet
📄 https://openreview.net/forum?id=xTJEN-ggl1b
📦 https://github.com/leaderj1001/LambdaNetworks
This media is not supported in your browser
VIEW IN TELEGRAM
PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models
📄 https://arxiv.org/pdf/2003.03808.pdf
📦 https://github.com/adamian98/pulse
▶️ https://colab.research.google.com/github/tg-bomze/Face-Depixelizer/blob/master/Face_Depixelizer_Eng.ipynb
📄 https://arxiv.org/pdf/2003.03808.pdf
📦 https://github.com/adamian98/pulse
▶️ https://colab.research.google.com/github/tg-bomze/Face-Depixelizer/blob/master/Face_Depixelizer_Eng.ipynb
CpZ_NOxy4ROlfDD2.mp4
6.1 MB
Taming Transformers for High-Resolution Image Synthesis
📄 https://arxiv.org/abs/2012.09841
📦 https://github.com/CompVis/taming-transformers
▶️ https://colab.research.google.com/github/CompVis/taming-transformers/blob/master/scripts/taming-transformers.ipynb
📄 https://arxiv.org/abs/2012.09841
📦 https://github.com/CompVis/taming-transformers
▶️ https://colab.research.google.com/github/CompVis/taming-transformers/blob/master/scripts/taming-transformers.ipynb
1 Trillion Parameters in new Language Model from Google
https://arxiv.org/abs/2101.03961
https://arxiv.org/abs/2101.03961