What is Machine Learning?
Machine learning (ML) is a field of computer science that uses statistical techniques to give computer systems the ability to “learn” (i.e., progressively improve performance on a specific task) with data, without being explicitly programmed.  From the above definition, three key words that pop into one’s mind are: data, computer systems, and statistical techniques. In the past decades, ML has been less popular largely because of limited amounts of accumulated data, low capabilities computer systems, and limited statistical methods for big data. Due to advancements in technology and research, there has been a tremendous increase in the amount of data being generated, a significant increase in computer systems’ capacities (both computational and storage), as well as the development of advanced statistical techniques that together portray ML as the main tool to drive digital enterprises. This is a reason why $47B is the estimated market value for machine learning by 2020. 
With rapid progress in technology come associated challenges, some of which include:
- Big data challenges: Big data is often described using five Vs; volume (amount of emails, tweets, photos, video clips, sensor data, etc.), velocity (speed at which social media, credit card, sensor data, etc. are generated), variety (structured, semi-structured, and unstructured data), veracity (messiness or trustworthiness of the data), and value (turning big data into business value). Each of these Vs has its challenges; for instance, in the past, we focused on structured data that neatly fit into relational databases but today, most data are unstructured and require different storage systems.
- Computer systems’ cost: The costs associated with computational power and storage are relatively low today compared to previous years, but the cost required to set up and maintain modern computer systems can be significant.
- Choice of technique: As stated by George E. P. Box, all models are wrong, but some are useful. As far as several advanced statistical techniques (algorithms) are concerned, this means skilled personnel are required to affect machine learning. This could be very costly given the limited number of such experts (data scientists).
Due to the numerous challenges associated with acquiring, storing, processing, and analyzing big data, deploying ML-embedded applications is costly, time consuming, and remains a tool that is used by tech/mega enterprises. SAP Leonardo Machine Learning comes in to facilitate the rapid deployment of ML embedded applications in all businesses by all enterprises, at very affordable rates.
Why SAP Leonardo?
Source: https://events.sap.com/sapandasug/en/session/32225, accessed April 13, 2018.
SAP Leonardo is the digital innovation system that enables enterprises to innovate at scale to confidently redefine their businesses.  SAP Leonardo is a cloud-based solution that provides a big data platform for the storage and retrieval of structured data in relational databases like SAP HANA and semi-structured/unstructured data in databases like Hadoop, with an interface to easily integrate data from the different sources for consumption by ML models and APIs. As a cloud-based solution comprising huge CPUs/GPUs and providing state-of-the-art speed and reliability, it serves as a “pay-as-you-go” system, meaning that all enterprises are now able to embed ML within their businesses without having to pay a fortune for computer systems. Additionally, SAP Leonardo ML comes with pre-trained routines that can easily be used by non-data scientists or customized (re-trained) to meet a specific business case. Apart from that, it allows for the importation of externally trained ML models from, for example, Google’s TensorFlow or R-for-rapid deployment and monitoring within the intelligent enterprise framework.
About this blog
This blog is intended to illustrate the power of SAP Leonardo ML through use cases. In the months ahead, we shall briefly touch on the following four main blocks of SAP Leonardo: IoT, Big Data, Analytics and ML but our focus shall be on ML. For details on these SAP Leonardo blocks, I refer you to their respective blogs elsewhere. Some of the use cases that we shall be covering include: document classification, topic detection, image classification, time series change-point detection, predictive maintenance, among others. We shall utilize pre-trained, customized, or external ML models where necessary. We welcome you to this exciting journey of intelligent enterprising with SAP Leonardo ML and are happy to have you aboard.
- Samuel, Arthur (1959). Some Studies in Machine Learning Using the Game of Checkers. IBM Journal of Research and Development, 3(3). doi:1147/rd.33.0210.
- IDC: https://www.idc.com/, accessed April 13, 2018.
- https://news.sap.com/africa/2017/05/17/introducing-the-new-sap-leonardo-empowering-companies-to-digitally-transform-at-scale/, accessed April 13, 2018.