A<\/b>rtificial Intelligence and Machine Learning have become more and more important. We are on the threshold to novel applications in industry and services, just as the recent success of AI projects has shown (Google\u2019s AlphaGo, IBM\u2019s Watson, autonomous cars). Four research groups at Freie Universit\u00e4t Berlin have now started the \u201cDahlem Center for Machine Learning and Robotics\u201d in order to explore machine learning algorithms and applications of intelligent systems. The following research groups are involved:<\/p>\n
Intelligent Systems and Robotics (Prof. Dr. Dr. habil. Ra\u00fal Rojas).<\/strong> Logic and automatic proofs (Prof. Dr. habil. Christoph Benzm\u00fcller).<\/strong> Autonomous Vehicles (Prof. Daniel Goehring).<\/strong> Artificial\u00a0and Collective Intelligence (Prof. Dr. Tim Landgraf).<\/strong> The Dahlem Center for Machine Learning and Robotics (DCMLR) offers specialized lectures, seminars and software projects that bring students closer to current research. Our courses include: pattern recognition\/machine learning, computer vision, image processing, robotics, software projects on autonomous vehicles, reinforcement learning and deep-learning, automated reasoning, artificial intelligence, and various seminars in all these fields.<\/p>\n For more information visit the following websites:<\/span><\/p>\n Automated Reasoning Leo III<\/a><\/p>\n AutoNOMOS Labs<\/a><\/span><\/p>\n Model Cars for Teaching at Universities<\/a><\/p>\n
\nProf. Rojas has been working on neural networks and machine learning since 1990. Neural networks are going through a Renaissance within the context of \u00a0\u201cDeep Learning\u201d methods. Project teaching is at the heart of the group\u2019s educational concept: students develop complete systems, for example, soccer robots that take part in the annual world championships. The group has been developing autonomous vehicles since 2006. The vehicles of the Lab are licensed for autonomous driving in Germany. For his combination of research and teaching Prof. Rojas was recognized as \u201cUniversity Teacher of the Year\u201d by the German Association of University Professors in 2014.<\/p>\n
\nModern formal systems enable computers to automatically search for proofs. Computers that interact with humans should not only show encyclopedic knowledge, but should also be able to argue logically.\u00a0This is the research focus of the Lab, where we have been developing automated reasoning systems, including the theorem prover LEO-II and his successor Leo-III. \u00a0LEO-II, a former world champion system, has recently been applied in theoretical philosophy\/metaphysics. It discovered relevant novel insights about G\u00f6del\u2019s ontological proof. Dr. Benzm\u00fcller received the 2016 central teaching award of FU Berlin for its interdisciplinary lecture course concept on computational metaphysics, which connects mathematics, computer science and philosophy. The integration of machine learning techniques has led to significant improvements in automated theorem proving, in particular with respect to the detection of relevant axioms in large knowledge bases.<\/p>\n
\nThis new Lab continues the development of autonomous vehicles at the FU Berlin. Prof. Goehring has developed algorithms for behavior and control of autonomous vehicles, and computer vision modules for the perception of the driving environment. Deep-learning on specific hardware is being applied to object recognition. The vehicle \u201cAutoNOMOS\u201d has a special permit for autonomous driving in Berlin. The car has been demonstrated on many occasions. Just recently, the car drove 2400 km in Mexico, from the border with Arizona to Mexico City. It was an impressive demonstration of the level of development of the FU-vehicles that are currently involved in BMBF and DFG projects.<\/p>\n
\nThe Landgraf Lab engages in interdisciplinary projects with biologists, physicists and neurobiologists, exploring mechanisms of collective intelligence. The team develops biomimetic robots in order to test hypotheses directly in interaction with the\u00a0biological system. It applies the latest machine learning methods for the analysis of high-dimensional data sets. Computer vision systems for the observation of social insects and fish schools have also been developed. The groups\u2019s robots can direct honeybee foragers via the bee dance and may also be used in water to examine the swarm behavior of fish. The Lab has also developed quadcopters for testing hypotheses about the neural correlates of navigation flying in insects.<\/p>\n