Machine Perception & Artificial Intelligence
Artificial Intelligence (AI) has a special role to play among the various topics included under the broad conceptual umbrella of the ‘digital transformation’.
On the one hand, this discipline has been in existence for over 60 years; on the other hand, it has experienced a huge push in recent years from new technologies such as Big Data and Cloud Computing. This year, it has really exploded. As an IT service provider with an integrated portfolio for the digital transformation, AI is a particularly exciting field for us, because it overlaps with our other subject areas at many points.
Artificial Intelligence: intelligent actions
Broadly speaking, Artificial Intelligence describes all digital systems designed to behave like a human being. In this context, human behaviour always refers to a specific, isolated task and not to imitation of overall human behaviour. If a system is designed to be able to perceive sections of reality, evaluate options, make decisions and trigger sensible actions like a human, we are dealing with an AI system.
Data science, Machine Learning and Deep Learning
Machine Perception is the ability of a computer system to interpret data as people do with all their senses. Generally, existing data first has to be made AI-capable, for example through annotation. AI thus often begins with a data management challenge before the creative part can get underway.
Big Data platforms, Deep Learning and Machine Learning are not synonyms for artificial intelligence, but rather core technologies for making machine perception and artificial intelligence possible and scalable.
Computer vision: object recognition
Our AI offering is closely connected to machine, algorithmic processing of sensory inputs – i.e. with the perception, analysis and further processing of spoken and written language, as well as images and videos. Example applications include the scanning and interpretation of vehicle registrations, receipts, etc. to improve the user experience or for quality assurance through computer vision (Industry 4.0). In the area of autonomous driving, AI solutions can classify objects using videos or images. There have also been initial investigations into how it might be possible to predict using CNNs (convolutional neural networks) whether a pedestrian wants to cross the road or not.
NLP: successful communication in text and speech
An example of a typical question regarding classifying content in texts is: how can content components from unstructured data such as contracts be filtered out without a human having to read the text? How can an AI solution create a knowledge graph from unstructured data? In the area of Search, we ask ourselves questions like: how can an AI solution find and represent knowledge, rather than just allowing a human to search for it? Or, in somewhat more technical terms: how can an AI application translate natural language and/or visual inputs into Elasticsearch queries? And from the users’ perspective: how can a chatbot in customer service provide support to ensure successful communication in text and speech?
Robotics: manipulation and movement
Other typical applications of Machine Perception and Artificial Intelligence are analyses of social media texts, sentiment analyses and AI-based data products such as recommender systems, voice assistants and voice APIs, as well as application implementations via robotic arms and robots. Humanoid robotics solutions such as Pepper and NAO are able to interact with people and can be used in retail, for example.
Reinforcement learning: learning through exploration
One method that has attracted attention in this area in recent years, for example in the shape of Google’s AlphaGo, is reinforcement learning. In this case, the algorithm learns through positive or negative feedback. In practical terms, reinforcement learning can be used whenever a simulation environment for the use case can be provided. Reinforcement learning is also used in NLP (natural language processing) when traditional optimisation processes are not applicable.
An interdisciplinary team of experts
inovex has built up a large team of proven experts in machine perception and artificial intelligence. Over 100 inovexperts now work in the area of data management and analytics, which also includes the enabling, adjoining technologies and methods. This breadth and depth of expertise allows us to turn demanding and complex tasks in these disciplines into productive solutions.
Some of our references:
Would you like a consultation on this subject?
I look forward to hearing from you!
Head of Data Management & Analytics
Our Technology Partners
inovex cooperates with a range of selected technology partners to offer our customers genuine added value: Amazon Web Services, Cloudera, Confluent, Elastic, e-shelter, Hortonworks, MapR, Microsoft, Quobyte and SoftBank Robotics.Read more