This is the first part I am writing about the journey of the data throughout the path of the prediction process in state-of-the-art algorithms. I am not sure if there is any same article like this; thus, I think it is the first kind of its own. So, I do like to keep writing such articles to see ML from a new viewpoint that data itself has. I also tried to make this article funny and story-like to be charming and understandable for the majority who are interested in Machine/Deep Learning. I’d consider it a baby, so please accept my apology for not being mature and please inform me of your suggestions to improve this type of article.

Anomalies are widespread when it comes to working on data. They become vital in time series. So, It is crucial to propose efficient methods to detect and deal with them. This article illustrates a state-of-the-art model called DGHL for anomaly detection. DGHL includes a ConvNet as a Generator and instead of encoding it maximizes the likelihood with the Alternating Back-Propagation algorithms. — As you may know, time series are everywhere, in any industry you are thinking of. …

These days, in deep learning, it is usual to hear about transformers’ outstanding performance on the challenges where other algorithms can not meet our expectations when most of them are based on attention. This article gives you a detailed illustration of the code and mathematics of the four most-used types of attention in the Deep Learning era. — The main feature of Attentions is the fact that their work is not limited to locality like CNNS, etc. but, we will see that in some cases we’d need the model to contemplate locality, etc. in their training process.

It is undeniable that when it comes to time-series forecasting, we need to forecast long dependencies for better decision-making in the future to cope with challenges regardless of the industry. Though transformers are revolutionary in the Deep Learning era, they contain some difficulties capturing long dependencies. As I discussed in the previous article ”Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting” about long dependencies to forecast the sequence length up to 480, we need algorithms beyond Transformers. This article is the same as the previous one but for longer sequence lengths which are highly demanded in industries. This article provides information about Autoformer (Decomposition Transformers with Auto-Correlation) to capture longer dependencies with an outstanding performance.

Time series are everywhere, in every industry from Energy to Geoscience, etc. Therefore, it is crucial to work on them; In most cases (especially in real-world projects), time-series datasets contain numerous missing data points which are highly connected to the output of prediction. This article gives you a review of the existed methods and then a thorough illustration of GRU-D (based on Gated Recurrent Unit) to deal with missing points.

This article is about proposing an advanced algorithm for forecasting time series and reducing high dimensionality. The researchers used auto-encoder architecture with convolutional operators to deal with the complexity of sequential data and amazingly they were successful. — About two years ago, I had to cope with time-series data and work on them. It was quite difficult for me at that time, because I hadn’t known anything about how to work on sequential data. Thus, I have started to read a large number of papers and finally, I…

Evolutionary Computation is a computational intelligence technique inspired by natural evolution. This method begins with the development of a group of people that respond to an issue; then, evaluate and modify the possible set of solutions to accomplish the best available solution. …

Informer is a transformer-based model that is developed to cope with long dependencies. The main topic of this article is sequence prediction. The sequence prediction can be found anywhere we have data that changes constantly, such as the stock market or etc. Despite most real-world applications of AI, predicting…

It is a great and admirable attempt to developed a single model to be able to work on multi tasks just like human vision. Exciting to see more advaces in computer vision. Instead of developing model architectures individually for the specific tasks (recognition of images, videos and, 3D data), in…