Invented by Suchitra Sundararaman, Jesper Lind, Juliy Broyda, Lev Sigal, Anton Ioffe, Yuri Arshavski, SAP SE

The market for machine learning for anomaly and fraud detection has been rapidly growing in recent years. With the increasing prevalence of digital transactions and the ever-evolving techniques used by fraudsters, businesses are turning to advanced technologies to protect themselves and their customers from financial losses and reputational damage. Machine learning, a subset of artificial intelligence, has emerged as a powerful tool in anomaly and fraud detection. It involves training algorithms to learn from large datasets and make predictions or take actions based on patterns and trends. This technology has the ability to analyze vast amounts of data in real-time, enabling businesses to detect anomalies and fraudulent activities quickly and accurately. One of the key advantages of machine learning in anomaly and fraud detection is its ability to adapt and improve over time. Traditional rule-based systems often struggle to keep up with the constantly evolving tactics employed by fraudsters. Machine learning algorithms, on the other hand, can continuously learn from new data and adjust their models accordingly. This adaptability allows businesses to stay one step ahead of fraudsters and detect previously unseen patterns or anomalies. The market for machine learning in anomaly and fraud detection is driven by several factors. Firstly, the increasing digitization of financial transactions has created a vast amount of data that needs to be analyzed. Machine learning algorithms excel at processing and analyzing large datasets, making them an ideal solution for fraud detection in the digital age. Secondly, the financial losses associated with fraud and the potential damage to a company’s reputation have made fraud detection a top priority for businesses across industries. Machine learning offers a more efficient and accurate approach to detecting anomalies and fraudulent activities, reducing the risk of financial losses and protecting a company’s brand image. Furthermore, regulatory requirements and compliance standards have become more stringent, necessitating robust fraud detection systems. Machine learning algorithms can help businesses meet these requirements by providing real-time monitoring and detection of suspicious activities, ensuring compliance with regulations and minimizing the risk of penalties or legal consequences. The market for machine learning in anomaly and fraud detection is also driven by the increasing availability of data and advancements in computing power. With the proliferation of connected devices and the Internet of Things (IoT), businesses have access to a wealth of data that can be leveraged for fraud detection. Additionally, the growing availability of cloud computing and powerful hardware accelerators has made it easier and more cost-effective to implement machine learning solutions. Several industries have already embraced machine learning for anomaly and fraud detection. The banking and financial services sector, in particular, has been at the forefront of adopting this technology. Banks and financial institutions are leveraging machine learning algorithms to detect fraudulent transactions, identify unusual patterns in customer behavior, and prevent identity theft. E-commerce platforms are also heavily reliant on machine learning for fraud detection. With the increasing volume of online transactions, these platforms face significant challenges in identifying and preventing fraudulent activities. Machine learning algorithms can analyze vast amounts of transactional data, identify suspicious patterns, and flag potentially fraudulent activities in real-time, protecting both the platform and its customers. In conclusion, the market for machine learning in anomaly and fraud detection is experiencing significant growth as businesses recognize the need for more advanced and efficient solutions to combat fraud. Machine learning algorithms offer the ability to analyze large datasets, adapt to evolving fraud techniques, and provide real-time detection and prevention of fraudulent activities. As the digital landscape continues to evolve, machine learning will play an increasingly crucial role in protecting businesses and their customers from financial losses and reputational damage.

The SAP SE invention works as follows

The present disclosure relates to systems, software and computer-implemented methods for auditing transactions. A method that is used in this example involves training a machine learning model on features which can be used to determine if an image is an original image of a document, or a document image automatically generated. This training takes place using both a set of images of documents and images of images of documents automatically generated. The request is made to classify the image as an authentic document image or an automatically created document image. Machine learning models are used to classify an image as an authentic document image or an automatically created document image based on the features in the image identified by the model(s). The image is classified. Machine learning models are updated according to the image and its classification.

Background for Machine learning for anomaly and fraud detection.

Travel and travel-related costs can be expensive for an organization. A travel expense management system automates the process of analyzing, monitoring, and controlling travel expenses and other reimbursable costs, all while increasing accuracy and worker productivity. A system that automates expense reporting can allow employees to spend less on creating and tracking expense reports. This allows them to focus more on their core jobs.

The present disclosure relates to systems, software and computer-implemented methods for expense reporting auditing. A method example includes: using a machine learning algorithm to identify features in a given image to determine if it is an authentic document image or an automatic generated document, the model identifies features in the image. The model then uses the features to classify that image.

Some or all aspects of this software may be implemented as computer-implemented methods, or included in systems or other devices to perform the described functionality. Details of these aspects and embodiments are described in the drawings and description below. The description, drawings and claims will reveal other features, advantages and objects of the disclosure.

DESCRIPTION of Drawings

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. 3A shows a timeline of the creation, submission and auditing of a expense report.

FIG. “FIG. 3B” illustrates a second timeline for the creation, submission and auditing of a expense report.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG. 9 is an example flowchart for secondary analysis after detecting a duplicate receipt.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG. 12 is an example flowchart for conducting a policy audit.

FIG. FIG. 13 shows an example of a system for auditing receipts.

Click here to view the patent on Google Patents.