Do like to stay informed yet sometimes struggle to skim through the heaps of articles? This Newsletter is an attempt to bring forward a blend of informative articles, recent news, and tweets from the Data World to your mailbox.
Welcome to the latest edition!
- The ultimate guide to starting AI
Cassie Kozyrkov, Chief Decision Scientist at Google breaks down the process of beginning your AI project step by step. The article emphasizes that the first step of any AI project is to form a clear vision of what you want from your AI/ML system i.e. to figure out desired outputs and objectives.
- Podcast: The Critical Art of Data Preparation
Data Scientists undeniably spend 80% of their time in collecting, cleaning, and preparing data that can be leveraged for analytics. In this podcast, Dean Abbott, a Data Science Leader shares his insight on "why data preparation is a critical art, and how the future of AI might be one with WAY more jobs than anyone anticipates."
- Facebook AI Research: Roadmap towards Machine Intelligence
Contributing to the history of Machine Intelligence, this paper proposed some fundamental properties these machines should have, focusing in particular on communication and learning and a concrete roadmap to develop them in realistic, small steps, that should lead to the ultimate goal of implementing a powerful AI.
- Top Evaluation Metrics For Your NLP Model
This article discusses some top evaluation metrics like NIST, BERT, SPICE and more that one should consider to capture the biases in their NLP model. "Even though learning biases has more to do with training data and less to do with model building, having a metric for capturing biases or a standard for biases would be a good practice to adopt."
- MIT's New Breast Cancer-Predicting shows Revolutionary Accuracy!
An AI system development called 'Mirai' by Prof. Regina Brazilay and graduate student Adam Yala can predict half of all incidences of breast cancer up to five years before they happen. “Mirai could transform how mammograms are used, open up a whole new world of testing and prevention, allow patients to avoid aggressive treatments, and even save the lives of countless people who get breast cancer.”
- Meta AI: The Artificial Intelligence Residency Program
Meta AI is offering a chance to maximize your experience in the field of AI by picking a research problem of mutual interest with the assigned team of AI Researchers then devising new Machine Learning Algorithms to solve it. "We encourage applications from people with technical backgrounds who hope to apply to a graduate program or would like more preparation before doing so."
- Standford Researchers' New Approach to Diagnostics
A new variation of MRI( Magnetic Resonance Imaging) called Quantitative MRI (qMRI) captures valuable non-visual metrics of the chemical structure compared to the traditional contrast between light, dark, and the grays revealing better insights into disease and damage than ever before. qMRI has yet to reach widespread use which can be helped by using AI to widen the use by more efficiently acquiring qMRI data."A promising technology is held back by lack of quality data, but with a newly released dataset, Stanford researchers are about to set it free."
- By @thomas_mock
#RStats tip of the day, want **reproducibly** random jitter plots? Use position_jitter()!— Tom Mock (@thomas_mock) December 21, 2021
ggplot(mtcars, aes(x = cyl, y = mpg, group = cyl)) +
geom_point(position = position_jitter(seed = 37))
Note the exact matching between plots!https://t.co/GaiuEpkUrN pic.twitter.com/eCL6RCZHCu
- By @mrogati
For most ML systems, data quality is *the* impact bottleneck. It sneaks up on you while the data team is busy modeling; when discovered it's dismissed as a 'bug, fixed' vs. a systemic issue.— Monica Rogati (@mrogati) October 28, 2021
Impressed with what @jeremystan and @eshmu built so far & their vision; congrats! https://t.co/QhbwLmSHCG
- By @GoogleAI
Today we present an open-source system to scale neural networks — often critical for improving model performance — by automatically parallelizing the model across devices, which enables researchers to more efficiently build and train large-scale models. https://t.co/FeSGtU5V26— Google AI (@GoogleAI) December 8, 2021
I hope you found today's edition informative. Thank You for the read. For more such newsletters - you can consider subscribing! :)