π¦π¦βοΈβοΈ
OpenAI trained and are open-sourcing a neural net called Whisper that approaches human level robustness and accuracy on English speech recognition.
Whisper is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitask supervised data collected from the web.
https://openai.com/blog/whisper/
Check out above link for paper, code and more details
π½π½
OpenAI trained and are open-sourcing a neural net called Whisper that approaches human level robustness and accuracy on English speech recognition.
Whisper is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitask supervised data collected from the web.
https://openai.com/blog/whisper/
Check out above link for paper, code and more details
π½π½
βοΈβοΈWhat is πππ₯π’ππ«πππ’π¨π§ π’π§ ππππ‘π’π§π ππππ«π§π’π§π ?
πCalibration is the property that tells us how well the estimated probabilities of a model match the actual probabilities, a.k.a the observed frequency of occurrences.
πCalibration can be represented using the Brier score. The Brier score is nothing more than the MSE between the actual and the estimated probabilities.
πThe two most common methods to address poor calibration is:
πplatt scaling and
πisotonic regression
πCalibration is the property that tells us how well the estimated probabilities of a model match the actual probabilities, a.k.a the observed frequency of occurrences.
πCalibration can be represented using the Brier score. The Brier score is nothing more than the MSE between the actual and the estimated probabilities.
πThe two most common methods to address poor calibration is:
πplatt scaling and
πisotonic regression
=======================================
βοΈβοΈVisual explanations of core machine learning concepts.
=======================================
πNothing can beat the use of infographics and interactivity when explaining some concept,
πFor Linear Regression
https://mlu-explain.github.io/linear-regression/
πFor all,
https://mlu-explain.github.io/
βοΈβοΈVisual explanations of core machine learning concepts.
=======================================
πNothing can beat the use of infographics and interactivity when explaining some concept,
πFor Linear Regression
https://mlu-explain.github.io/linear-regression/
πFor all,
https://mlu-explain.github.io/
This media is not supported in your browser
VIEW IN TELEGRAM
π«₯π«₯FLAG: Flow-based 3D Avatar Generation from Sparse Observations
πPaper :
https://microsoft.github.io/flag/files/paper.pdf
πLink :
https://www.microsoft.com/en-us/research/publication/flag-flow-based-3d-avatar-generation-from-sparse-observations/
πPaper :
https://microsoft.github.io/flag/files/paper.pdf
πLink :
https://www.microsoft.com/en-us/research/publication/flag-flow-based-3d-avatar-generation-from-sparse-observations/
𧬠The data structure for unstructured multimodal data · Neural Search · Vector Search · Document Store
For doc
https://docarray.jina.ai/
For GitHub
https://github.com/jina-ai/docarray
For doc
https://docarray.jina.ai/
For GitHub
https://github.com/jina-ai/docarray
Transformers in Time Series: A Survey
A curated list of awesome resources (papers, code, data) on Transformers in Time Series categorized by tasks, including:
β’ Forecasting
β’ Anomaly detection
β’ Classification
Transformers capture long-range dependencies and interactions.
abs: https://arxiv.org/abs/2202.07125
pdf: https://arxiv.org/pdf/2202.07125.pdf
Awesome list repo: https://github.com/qingsongedu/time-series-transformers-review
A curated list of awesome resources (papers, code, data) on Transformers in Time Series categorized by tasks, including:
β’ Forecasting
β’ Anomaly detection
β’ Classification
Transformers capture long-range dependencies and interactions.
abs: https://arxiv.org/abs/2202.07125
pdf: https://arxiv.org/pdf/2202.07125.pdf
Awesome list repo: https://github.com/qingsongedu/time-series-transformers-review
googlefinance
Python module to get stock data from Google Finance API. This module provides no delay, real time stock data in NYSE & NASDAQ.
$pip install googlefinance
https://github.com/hongtaocai/googlefinance
Python module to get stock data from Google Finance API. This module provides no delay, real time stock data in NYSE & NASDAQ.
$pip install googlefinance
https://github.com/hongtaocai/googlefinance
GitHub
GitHub - hongtaocai/googlefinance: Python module to get real-time stock data from Google Finance API
Python module to get real-time stock data from Google Finance API - hongtaocai/googlefinance
Monday, 26 September 2022
Latest AI Curated Track
βοΈβοΈA third of scientists working on AI say it could cause global disaster. A survey covering the opinions of 327 researchers who had recently co-authored papers on AI research in natural language processing.
βοΈβοΈTeslaβs AI Day 2022 is scheduled for September 30 in Palo Alto, California.
βοΈβοΈAI will help phone photos surpass the DSLR, says Qualcomm
βοΈβοΈOver the past few weeks, researchers at Google have demoed an AI system, PaLI, that can perform many tasks in over 100 languages
βοΈβοΈA Berlin-based group launched a project called Source+ that's designed as a way of allowing artists, including visual artists, musicians and writers, to opt into and out of allowing their work being used as training data for AI.
βοΈβοΈAccording to International Data Corp, China will invest US$26.7 billion in artificial intelligence in 2026. With regards A total of 45,000 AI-related patent applications were filed in Shanghai
βοΈβοΈSaudi Arabia focuses on AI-driven economy, considers data the new oil: SDAIA
βοΈβοΈAn artist based in New York City has been granted the first known registered copyright for artwork made using latent diffusion AI.
βοΈβοΈSalesforce AI Open-Sources βLAVIS,β A Deep Learning Library For Language-Vision Research/Applications.
βοΈβοΈHarvard celebrated the launch of the Kempner Institute for the Study of Natural and Artificial Intelligence on Thursday
βοΈβοΈCleareye AI Announces Strategic Alliance with J.P. Morgan
βοΈβοΈAI proves to be more accurate in diagnosing cardiac function than sonographers
βοΈβοΈAmazon SageMaker Provides New Built-in TensorFlow Image Classification Algorithms available in tensor flow hub.
βοΈβοΈ Google today released TensorFlow Graph Neural Networks (TF-GNN) in alpha, a library designed to make it easier to work with graph structured data using TensorFlow, its machine learning framework.
Inscribed by,
Raja
Latest AI Curated Track
βοΈβοΈA third of scientists working on AI say it could cause global disaster. A survey covering the opinions of 327 researchers who had recently co-authored papers on AI research in natural language processing.
βοΈβοΈTeslaβs AI Day 2022 is scheduled for September 30 in Palo Alto, California.
βοΈβοΈAI will help phone photos surpass the DSLR, says Qualcomm
βοΈβοΈOver the past few weeks, researchers at Google have demoed an AI system, PaLI, that can perform many tasks in over 100 languages
βοΈβοΈA Berlin-based group launched a project called Source+ that's designed as a way of allowing artists, including visual artists, musicians and writers, to opt into and out of allowing their work being used as training data for AI.
βοΈβοΈAccording to International Data Corp, China will invest US$26.7 billion in artificial intelligence in 2026. With regards A total of 45,000 AI-related patent applications were filed in Shanghai
βοΈβοΈSaudi Arabia focuses on AI-driven economy, considers data the new oil: SDAIA
βοΈβοΈAn artist based in New York City has been granted the first known registered copyright for artwork made using latent diffusion AI.
βοΈβοΈSalesforce AI Open-Sources βLAVIS,β A Deep Learning Library For Language-Vision Research/Applications.
βοΈβοΈHarvard celebrated the launch of the Kempner Institute for the Study of Natural and Artificial Intelligence on Thursday
βοΈβοΈCleareye AI Announces Strategic Alliance with J.P. Morgan
βοΈβοΈAI proves to be more accurate in diagnosing cardiac function than sonographers
βοΈβοΈAmazon SageMaker Provides New Built-in TensorFlow Image Classification Algorithms available in tensor flow hub.
βοΈβοΈ Google today released TensorFlow Graph Neural Networks (TF-GNN) in alpha, a library designed to make it easier to work with graph structured data using TensorFlow, its machine learning framework.
Inscribed by,
Raja
AWS Question: A customer has a workload that will run for total of 6 months and can withstand interruptions. What would be the most cost-efficient Amazon EC2 instance purchasing option?
Anonymous Quiz
56%
On-Demand Instance
22%
Spot instance
0%
Dedicated Instance
22%
Reserved Instance
π£ Meta AI:
π«₯ Make-a-video: The text-to-video generation without text-video data
π«₯ Paper: https://makeavideo.studio/Make-A-Video.pdf
π«₯ Project page: makeavideo.studio
An effective method that extends a diffusion-based T2I model to T2V through a spatiotemporally factorized diffusion model.
π«₯ Make-a-video: The text-to-video generation without text-video data
π«₯ Paper: https://makeavideo.studio/Make-A-Video.pdf
π«₯ Project page: makeavideo.studio
An effective method that extends a diffusion-based T2I model to T2V through a spatiotemporally factorized diffusion model.
π£π£An AI used medical notes to teach itself to spot disease on chest x-rays
π«₯π«₯A team of researchers from Harvard Medical School trained the CheXzero model on a publicly available data set of more than 377,000 chest x-rays and more than 227,000 corresponding clinical reports. This taught it to associate certain types of images with their existing notes, rather than learning from structured data that had been manually labeled for the task.
Paper:
https://www.nature.com/articles/s41551-022-00936-9
News link:
https://www-technologyreview-com.cdn.ampproject.org/c/s/www.technologyreview.com/2022/09/15/1059541/ai-medical-notes-teach-itself-spot-disease-chest-x-rays/amp/
π«₯π«₯A team of researchers from Harvard Medical School trained the CheXzero model on a publicly available data set of more than 377,000 chest x-rays and more than 227,000 corresponding clinical reports. This taught it to associate certain types of images with their existing notes, rather than learning from structured data that had been manually labeled for the task.
Paper:
https://www.nature.com/articles/s41551-022-00936-9
News link:
https://www-technologyreview-com.cdn.ampproject.org/c/s/www.technologyreview.com/2022/09/15/1059541/ai-medical-notes-teach-itself-spot-disease-chest-x-rays/amp/
π£CLIP: The Most Influential AI Model From OpenAI β And How To Use It
π CLIP stands for Constastive Language-Image Pretraining
π CLIP is an open source, multi-modal, zero-shot model.
π Given an image and text descriptions, the model can predict the most relevant text description for that image, without optimizing for a particular task.
πCLIP is trained using a staggering amount of 400 million image-text pairs. For comparison, the ImageNet dataset contains 1.2 million images.
πThe final tuned CLIP model was trained on 256 V100 GPUs for two weeks. For an on-demand training on AWS Sagemaker, this would cost at least 200k dollars!
π The model uses a minibatch of 32,768 images for training.
π CLIP stands for Constastive Language-Image Pretraining
π CLIP is an open source, multi-modal, zero-shot model.
π Given an image and text descriptions, the model can predict the most relevant text description for that image, without optimizing for a particular task.
πCLIP is trained using a staggering amount of 400 million image-text pairs. For comparison, the ImageNet dataset contains 1.2 million images.
πThe final tuned CLIP model was trained on 256 V100 GPUs for two weeks. For an on-demand training on AWS Sagemaker, this would cost at least 200k dollars!
π The model uses a minibatch of 32,768 images for training.
π£π£Daft :
A fast, ergonomic and scalable open-source dataframe library: built for Python and Complex Data/Machine Learning workloads.
π«₯Install
pip install getdaft.
π«₯Why Daft ?
Processing Complex Data such as images/audio/pointclouds often requires accelerated compute for geometric or machine learning algorithms.
π«₯Official Link
https://getdaft.io/
A fast, ergonomic and scalable open-source dataframe library: built for Python and Complex Data/Machine Learning workloads.
π«₯Install
pip install getdaft.
π«₯Why Daft ?
Processing Complex Data such as images/audio/pointclouds often requires accelerated compute for geometric or machine learning algorithms.
π«₯Official Link
https://getdaft.io/