Deep learning has grown its popularity as we have more access to data, powerful computation, and advancements in algorithms. Companies have used deep learning to solve labor intensive problems and develop new technologies never before seen. In this section, we will cover six applications of deep learning that can be seen today.
Self driving cars have been under development over the last 100 years with the first semi-autonomous car experiment taking place in the 1920's. Since then, there have been many different experiments done. However, only with advancements in deep learning technology in the past few years, have we seen successful autonomous driving cars. One of the companies at the forefront of self-driving cars is Tesla. Tesla has used deep learning neural networks trained with raw data taken from an array of sensors and cameras on their cars to train neural networks with 48 networks that took 70,000 GPU hours to train. Their deep learning network outputs 1,000 distinct predictors to analyze at each step. Using deep learning, Tesla is able to create autonomous driving vehicles that are continuously learning while on the road.
The healthcare industry has many different applications for deep learning however one of the most prevalent uses is for image analytics and diagnostics. Using convolutional neural networks the healthcare industry has better been able to analyze images from x-rays and MRIs. By training deep learning neural networks with images from x-rays and MRIs showing positive diagnosis's, some neural networks have shown the same accuracy as industry professionals and in some cases more accurate. In a study published in the Annals of Oncology deep learning convolutional neural networks trained to identify melanoma from images were 10% more accurate in positively identifying melanoma than trained human clinicians.
Data collected on users based on their search history, social media, and online tendencies has allowed for better prediction than ever before of a person's interests. Advertising companies can use deep learning to provide targeted advertising for users based on their interests but can also predict the types of advertisements they might be interested in based on the data gathered from people with similar traits. Advertising companies that use deep learning are able to increase their revenue and reduce costs by analyzing and taking action on data on a large scale without direct input. They are also able to gain a competitive advantage over other companies by having a greater prediction capabilities and better speed than the competition.
Another common use of deep learning technology is for Natural Language Processing. Natural language processing is converting converting speech into text. Natural language processing is done by training neural networks with audio files and transcribed text to build neural networks that can accurately and efficiency convert speech to text. This technology is used across many different industries and has many different applications. One of the most profound applications is in voice recognition software like the voice recognition software used by Amazon's Alexa. Amazon sends the audio of a request made to an Alexa to their servers where is it converted to text and analyzed using their deep learning algorithms.
Companies use deep learning to better improve their customer service experience. Using chat bot's has allowed companies to save resources while still offering customers the same customer service experience. Customer service chat bots are deep learning algorithms trained using text from past customer service experiences to answer questions or solve customers problems without having to wait to speak to a representative.
Deep learning has been used in detecting finical and credit card fraud. Training neural networks with data depicting finical fraud has allowed for deep learning algorithms that can detect fraud in real time. Rather than having people spend time going over finical transactions, deep learning algorithms can detect potential cases of fraud by going through lager amounts of data all at once. This algorithms can also detect fraud that may have been missed by a human by detecting very subtle predictors.
- https://www.tesla.com/autopilotAI
- https://healthitanalytics.com/features/what-is-deep-learning-and-how-will-it-change-healthcare
- https://www.marketingaiinstitute.com/blog/ai-in-advertising
- https://www.wired.com/story/amazon-alexa-2018-machine-learning/
- https://hackernoon.com/deep-learning-chatbot-everything-you-need-to-know-r11jm30bc
- https://www.altexsoft.com/whitepapers/fraud-detection-how-machine-learning-systems-help-reveal-scams-in-fintech-healthcare-and-ecommerce/