There is a paper recently published to show the Neural Networks are never perfect and being easily fooled. In the paper, the research team is giving “Strange Poses of Familiar Objects” for a deep neural network engine to recognize. However, the result of detection is extremely bad with more than 99% the poses are mis-classified.
Here is the original paper published:
In reality, the collection of data may lead to “Strange Poses” and not a perfect data. Thus, the writers of the paper suggested 3D datasets / objects to study rather than just 2D image to reduce the chance of classification failures.
In this week, we would like to share an article with highlighting the keys of success for Machine Learning.
For the original article, the author(s) had list out the keys of success for machine learning:
- Start small with Machine Learning – this is similar to all other data analytic projects to have a smaller scoped for better management
- Machine Learning Must Have Data Quality to Succeed – data is the most important ingredient for data analytic; so, the quality is critical.
- No Universal Machine Learning Algorithms – Machine Learning itself is the approach to solve one single “specific” problem and the related algorithms should be unique by the corresponding use-case.
Original Article at DataVersity.net
In this week, we would like to share an article discussing the basic concepts of AI with the role and importance of training. Training absolutely crucial that everyone involved in the development of your model understands how it works.
Original Article: https://insidebigdata.com/2018/10/08/ai-training-work/
The Arthur for the article uses an example for a basket of songs in English and Spanish for the machine to categorize the song according to the language. He points out a number of common problems, such as:
- Data quality for training
- Overfitting of Sample
- Testing considerations
Personally, he is discussing machine learning – a subset of AI rather than the huge picture of AI. Nevertheless, he has illustrated the fundamentals of the “machine learning” processing steps and concerns. It is still an article worth to read for beginners.