Alexis Networks

Alexis Networks
Understanding Anomaly Detection

Wednesday, June 17, 2020

Anomaly Detection by Aashrith Sai

 

Aashrith Sai


 
About Myself
Currently pursuing my masters with keen focus on Deep learning and data science. I am passionate about data at every stage of the data life-cycle in a project. I have a profound understanding about several classification and regression algorithms with comprehensive understanding of data science project pipeline, feature engineering, Data Visualization, Data-Modeling and Statistics. I am currently working as a Deep Learning Intern at Alexis Networks and developing AI solutions for human anomalies by tracking their actions. 

Linked-In       |         Email Me


Anomaly detection by Aashrith Sai
According to me anomaly is anything that is deviating from what is expected or normal or standard. In every organization or a business, everything is planned to meet the expectations of the clients or consumers. In all these scenarios one must be aware of the anomalies to successfully execute their plan. The more we know about the probability of an event happening in the future, the more prepared we can be in the present to handle the situation in future.


Real-Time Scenario: Let us look at an example of a grocery store.

Fig1: Every weekend there are a lot of customers.

Fig 2: The weekend when there was a thunderstorm.
                                                                     

As the number of customers are very high on the weekends the grocery store owner hires more staff during the weekends and purchases more goods from the vendors. The owner would have saved a lot of money if he had a prior idea of the thunderstorm affecting his sales. In this case a thunderstorm is an anomaly. Lot of resources can be saved and plans can be executed more efficiently if we have the information about the anomalies happening in future, present and past.


Internship experience at Alexis:

Alexis Networks is the first company I am working with in the US. Through the last few weeks I have been given an opportunity to explore a lot of new technologies in the field of machine learning, cloud and many more that are essential for product development. Through this opportunity, I have discovered my new interests towards Amazon AWS which is driving me to work on getting certified as an AWS machine learning developer. I am thoroughly enjoying the freedom given to me to explore new paths to have a complete learning experience.
My excitement towards new projects, products and tasks reasserted the fact that I am in the right direction of exploring technology and complementing my skill-set to become successful in my career. 


What have you learned thus far compared to what you initially expected?

I always wanted to work on products that would impact people to make their life more easier and comfortable. At Alexis, I am working on exactly what I wanted to do. I wanted to get a real time experience with deep learning, which I have developed in the initial few weeks of my internship to detect human anomalies. I have been working and learning more about AWS cloud for machine learning and deep learning applications. I am learning how to package and deliver products to the clients from the stage of development to deployment. I am really enjoying this process and never expected that this opportunity would open up so many paths to learn and explore.

My cough and sneeze detection model:



What has your experience been like during COVID19 and how has it impacted you?

It is always good to have people around you and discuss new ideas and share each other's experiences. I am glad that though this situation is keeping us all at our home, we can still connect with each other and discuss the blockers using social media and virtual meets. I am really happy on how everyone is available and contributing to make my internship experience complete.



How insurers can use big data to manage the COVID-19 pandemic

Caught between higher demand and tightening supply, insurers are looking to technology to compute adequate rates and come up with new types ...