Модульбанк (http://vk.com/club80328935) дает советы и делится практическим опытом подготовки данных
"В задачах машинного обучения качество моделей очень сильно зависит от данных.
Но сами данные в реальных задачах редко бывают идеальными. Как правило, самих данных не много, количество доступных для анализа параметров ограничено, в данных шумы и пропуски. Но решать задачу как-то нужно..."
https://habrahabr.ru/post/342366/
"В задачах машинного обучения качество моделей очень сильно зависит от данных.
Но сами данные в реальных задачах редко бывают идеальными. Как правило, самих данных не много, количество доступных для анализа параметров ограничено, в данных шумы и пропуски. Но решать задачу как-то нужно..."
https://habrahabr.ru/post/342366/
Google выпустила картонную камеру за $45. Она поддерживает нейросети
https://republic.ru/posts/88098
https://blog.google/topics/machine-learning/introducing-aiy-vision-kit-make-devices-see/
https://republic.ru/posts/88098
https://blog.google/topics/machine-learning/introducing-aiy-vision-kit-make-devices-see/
Amazon DeepLens — первая Wi-Fi камера с Deep Learning
https://geektimes.ru/company/intel/blog/296201/
https://geektimes.ru/company/intel/blog/296201/
IBM представила первые серверы Power9 для ИИ
https://hightech.fm/2017/12/07/power9
https://hightech.fm/2017/12/07/power9
ИИ от Google за несколько часов сам научился играть в шахматы и го и победил «чемпионов»
https://rb.ru/news/alpha-zero/
https://arxiv.org/pdf/1712.01815.pdf
https://rb.ru/news/alpha-zero/
https://arxiv.org/pdf/1712.01815.pdf
Google представила Tacotron 2, новую систему для преобразования текста в речь
https://tproger.ru/articles/tacotron-2-text-into-speech/
https://tproger.ru/articles/tacotron-2-text-into-speech/
Stacking Made Easy: An Introduction to StackNet by Competitions Grandmaster Marios Michailidis (KazAnova)
http://blog.kaggle.com/2017/06/15/stacking-made-easy-an-introduction-to-stacknet-by-competitions-grandmaster-marios-michailidis-kazanova/
http://blog.kaggle.com/2017/06/15/stacking-made-easy-an-introduction-to-stacknet-by-competitions-grandmaster-marios-michailidis-kazanova/
Асинхронная загрузка больших датасетов в Tensorflow
https://habrahabr.ru/post/345546/
https://habrahabr.ru/post/345546/
https://github.com/sld/convai-bot-1337
Skill-based Conversational Agent that took 1st place at 2017 NIPS Conversational Intelligence Challenge (http://convai.io)
The chatbot developed for the ConvAI challenge. Our bot is capable of conversing with humans about given text (e.g. a paragraph from Wikipedia article). The conversation is enabled by a set of skills, including chit-chat, topics detection, text summarization, question answering and question generation. The system has been trained in a supervised fashion to select an appropriate skill for generating a response. Furthermore, we have developed an overall dialog quality scorer and next utterance scorer to correct agent's policy. Our bot is implemented with open source software and open data; it is self-hosted, and employs a supervised dialog manager with a linear hierarchy. The latter allows a researcher to focus on skill implementation rather than finite state machine development.
Skill-based Conversational Agent that took 1st place at 2017 NIPS Conversational Intelligence Challenge (http://convai.io)
The chatbot developed for the ConvAI challenge. Our bot is capable of conversing with humans about given text (e.g. a paragraph from Wikipedia article). The conversation is enabled by a set of skills, including chit-chat, topics detection, text summarization, question answering and question generation. The system has been trained in a supervised fashion to select an appropriate skill for generating a response. Furthermore, we have developed an overall dialog quality scorer and next utterance scorer to correct agent's policy. Our bot is implemented with open source software and open data; it is self-hosted, and employs a supervised dialog manager with a linear hierarchy. The latter allows a researcher to focus on skill implementation rather than finite state machine development.
AI in 2018 for developers
Hi again! In last article I tried to show my vision on what research areas are maturing and can grow big this year. Research is cool, but there must be something from AI world that became mature in 2017 and is ready now to be used in mass applications. This is what this article will be about — I would like to tell about technologies that are good enough to be used in your current work or to build your own startup based on them.
Alexandr Gonchar (http://vk.com/id20143288)
https://medium.com/@alexrachnog/ai-in-2018-for-developers-2f01250d17c
Hi again! In last article I tried to show my vision on what research areas are maturing and can grow big this year. Research is cool, but there must be something from AI world that became mature in 2017 and is ready now to be used in mass applications. This is what this article will be about — I would like to tell about technologies that are good enough to be used in your current work or to build your own startup based on them.
Alexandr Gonchar (http://vk.com/id20143288)
https://medium.com/@alexrachnog/ai-in-2018-for-developers-2f01250d17c
The New Version Of Tensorflow 1.5.0 Released! (Выпущена новая версия Tensorflow 1.5.0!)
#DeepLearning #Tensorflow
https://www.techleer.com/articles/457-the-new-version-of-tensorflow-150-released/
#DeepLearning #Tensorflow
https://www.techleer.com/articles/457-the-new-version-of-tensorflow-150-released/