Article

Deloitte Digital | Eight annual AI achievements in 2018 – Deloitte Analytics Institute (DAI)

Innovation with 'Digital Difference'

To lead the exponential transformation in our digital age, Deloitte has launched "Digital Difference" in 2018. We are committed to cutting edge approaches that help businesses respond to technological disruption and give us an unparalleled advantage over our peers. In response to the digital trends and corporate cultural transformation, Deloitte China has made remarkable achievements following our "Think. Create. Do." transformation methodology fuelled by "Digital Difference".

AI (Artificial Intelligence) has become a powerful engine for digital transformation. In 2018, DAI has enhanced its technical strength by diving deep into  NLP/Chatbot/face recognition/speech recognition/image recognition/OCR/ machine learning/RCA and related areas. DAI wisely used these technologies to help various functions develop innovative AI platform, smart hearing system, BLAST and other intelligent products. In the meanwhile, DAI has collaborated with external AI Partners (SenseTime, iFlytek, etc.) to help clients adapt and thrive – becoming the disruptors, not the disrupted – well into the future.

 

Eight AI achievements

1. Deloitte AI chatbot- DAI joins hands with TMC to support the health care industry

With the development of natural language technology, "Deloitte AI chatbot" can effectively help share workload regarding corporate internal information collection and enquiry, therefore significantly reducing manual labor. Compared with traditional search engine, the chatbot, powered by dialogue technology, can achieve 24-hour customer service, answering questions anytime and anywhere. This system also provides a variety of deployment solutions tailored to clients' needs.

Deloitte AI chatbot, which incorporates various cutting-edge AI technologies such as natural language processing (NLP), knowledge graphs, deep learning and multi-agent conversations, etc., is the first AI chatbot of Deloitte China. Its competitive advantage lies in the ability to take on more online tasks, such as analysis of questions asked, rapid process approval, and one-click calls, etc. DAI has successfully developed a customized chatbot with TMC team for Herbalife, which has broadened Deloitte's service scope in the AI era and guided the way for enterprise innovation and digital transformation. More importantly, it helped enterprises improve operational efficiency and take the lead in the AI era.

How Deloitte AI chatbot works

 

2. Deloitte AI platform helps pharmaceutical companies predict the future

DAI has worked with the Assurance-DIFA team to help a well-known pharmaceutical company forecast sales via "Deloitte AI platform". This is first time that Deloitte used AutoML-based forecast platform to make predictions. The platform incorporates techniques such as time series forecasting, feature models, interpretability and data visualization. With the advantages of high accuracy, interpretability and user friendliness, Deloitte AI platform can analyse project data from the business and technical aspects respectively, with highly precise results. There is a wide range of possible application scenarios for such smart forecast system. For example, sales prediction can be applied in most sales-related industries, as well as industries involving warehousing management.

Prediction by AI platform

 

3. AI prediction for Russian World Cup

There is a wide range of application scenarios of AI technologies in prediction. During every World Cup game, AI companies around the world would allocate the best scientist team to predict results with leading technologies, in order to highlight their technical capabilities. During the 2018 World Cup in Russia, DAI of Deloitte China leveraged big data to predict results, aligned with our "Think. Create. Do." principle. It is the first time that DAI leveraged AI technologies such as complex graph network and data visualization to predict the results of a world sports event. By the end of the World Cup, DAI forecasted the results in the last 13 days with 85% accuracy in average through its original "Anti-Interoperability Network Algorithm" and millions of model evaluations.

Complex network of AI prediction for World Cup

 

4. DAI works with IDDC in incubating smart AI hearing platform for an automobile company

Deloitte smart AI hearing platform has successfully converted the telephone or sales recording into texts, which serve as the basis for next-step keyword recognition, speech analysis and staff evaluation. Then the company can make improvements accordingly based on the evaluation. At the same time, the company can create customer portraits according to the dialogue, thereby initiating a personalized marketing program to achieve intelligent and accurate marketing. DAI closely collaborated with IDDC on in-depth use of voice-to-text, natural language processing, and big data analysis, etc., in order to create a voice-to-text recording analysis platform. Smart AI hearing platform is widely used in automotive, real estate, e-commerce, banking, telecommunications and other industries, aiming at helping clients expand new technology-driven services, as well as guiding the way for corporate innovation and digital transformation.

Analysis result of smart AI hearing platform

 

5. OCR evaluation report brings about new intelligent 

DAI has created a scientific, objective and comprehensive evaluation report for some selected mainstream OCR products from various dimensions including format compatibility, accuracy, stability, and user friendliness, etc., focusing on their performance in identifying key phrases in diverse documents formats such as tables, and invoices. Leveraging optical character recognition technology and deep learning algorithm, OCR products are mainly applied in the firm's audit & assurance, taxation and other professional services, to promote efficient and intelligent input of information. After releasing the evaluation report, various functions/teams applied OCR technology to their products and projects. DAI has also utilised OCR technology to assist internal teams in prototype verification and development.

Evaluation results of multi-format invoice OCR

 

6. DAI private cloud stimulates infinite business imagination with face recognition prototype

Deloitte face recognition prototype system, powered by HOG and deep learning, utilises both CPU and GPU resources. Based on online open-source projects, it enables image-based face verification and recognition, video-based face recognition, face recognition with multiple images and videos, and helps determine whether pairs of face images showed the same person. The face recognition prototype system incorporates RCNN, CNN, Siamese network and other technologies to explore the possibility for automated image recognition that used to require a lot of manual participation.

Face recognition prototype

 

7. RPA upgrade: From RPA to RCA

RPA refers to Robotic Process Automation technology. Against the background of intellectualization, traditional RPA can no longer fully address the demand of clients. Incorporating artificial intelligence technologies (such as verification code recognition, OCR, NLP, etc.) into traditional RPA, DAI has empowered the traditional RPA and upgraded it into Robotic Cognitive Automation (RCA). Through in-depth cooperation between DAI and the Assurance DIFA team, the firm has extended and applied the latest RCA technology in a number of finance projects in the banking industry, laying a solid foundation for more RCA projects in the future.

Required technologies for RPA upgrade

 

8. Bank Loan Analysis SysTem (BLAST) Prototype

The Bank Loan Analysis SysTem (BLAST), jointly developed by DAI and Audit FSI team, is a comprehensive perspective analysis system specifically designed for bank loan business. It leverages years of experience, understanding and prospective analysis of the banking industry from Deloitte experts, with deep integration of industry-leading visual analytics technology and AI machine learning algorithm. BLAST is intelligent, easy to use and integrate, and well capable of conducting real-time analysis. It can adapt to the explosive growth of demand for banking business data, quickly collect and analyse data, and update the data information in real time. Combining the advantages of data warehouse and real-time data processing, the system aims to help solve the most complex business problems for banking clients utilising visual analytics and intelligent analysis technology.

Functions of the Bank Loan Analysis SysTem (BLAST)

The prospect for 2019

In 2019, DAI will continue to uphold the principle of "Think. Create. Do" in driving transformation, committed to helping clients adapt and thrive – becoming the disruptors, not the disrupted – well into the future. DAI welcomes you to actively make contact and work together with us to create a better Deloitte and a better future.

Did you find this useful?