Deep learning will be faster and data collection better
Nowadays, deep learning faces certain challenges associated with the data collection and the complexity of the computations. Innovations in hardware are now being developed to speed up the deep learning experiments, e.g. the new GPUs with a greater number of cores and a different form of architecture. According to Marc Edgar, a Senior Information Scientist at the GE Research, deep training will shorten the development time of software solutions from several months to several days within the next 3-5 years. This will improve the functional characteristics, increase productivity, and reduce product costs.
Currently, most large firms realize the importance of data collection and its influence on the business effectiveness. In the coming year, companies will start using even more data, and the success will depend on the ability to combine the disparate data. In 2018, companies will collect customer data via CRM, ticket systems, BMP and DMP, as well as the omnichannel platforms. The popularity of collecting data on specialized sensors like LIDAR is also on the rise. Integrating the existing systems with all types of client data into a single information pool will definitely be on trend. Startups will continue to create new methods for gathering and using data, further reducing the costs.
