Difference: MiOverview (r3 vs. r2)

The Machine Intelligence Laboratory (formerly known as the Speech, Vision and Robotics group) was founded by the late Professor Frank Fallside in the early 1970's, when the main interests were in speech processing and control applications. In the mid 1980's, the laboratory developed a strong interest in the theory and application of neural networks and this led to a widening of the laboratory's research to include vision and robotics. Medical imaging followed in the mid 1990's. Today, the guiding principle of all research in the laboratory is that a well-designed engineering system must be based on a sound mathematical model. In this regard, neural networks represent just one of a wide range of applicable techniques. Others include stochastic processes such as hidden Markov models, Bayesian inference, invariant transformations in 3D geometry, computational geometry, Wiener and Kalman filtering, classification and regression trees, and genetic algorithms.

• Neural networks, pattern recognition and machine learning, including multi-layer perceptrons, radial basis functions, and recurrent networks.
• Signal processing, non-stationary time-series analysis, speech coding and compression.
• Speech recognition using both neural networks and hidden Markov Models. This includes large vocabulary recognition, recognition in noise, speaker adaptation and word spotting.
• Language processing including N-grams, stochastic context-free grammars, grammatical inference, dictionary construction.
• Statistical machine translation
• Image processing and object recognition, including 3-D reconstruction from 2-D images, image segmentation, and face recognition.
• Visual navigation of mobile robots and task level and sensor-based robot control within an unstructured environment.
• Aspects of robot assembly including path planning, hand-eye coordination and quality inspection using computer vision, man-machine interfaces using visual gestures.
• Aspects of medical imaging, including the acquisition, visualisation, registration and segmentation of 3D ultrasound images for medical diagnosis.
• Risk analysis in various aspects of health care.

In addition to supporting a large post-graduate research activity, the Machine Intelligence Laboratory is also responsible for a Masters course in Advanced Computer Science and undergraduate teaching in the areas of computing and pattern processing. The Masters course is a one year course run jointly with the Computer Laboratory. It has an annual enrolment of around 20 students and its aim is to teach both the theory and practice of speech and language processing systems. Topics covered include speech analysis, recognition and synthesis; syntax and parsing; semantics and discourse analysis; and perception and psycholinguistics. The course consists of two terms of taught lectures and practicals followed by a three month thesis project. It operates with the support of EPSRC and it has close links with UK industry via an industrial advisory board.

At the undergraduate level, the laboratory is involved in the teaching of information engineering and computing generally. It is responsible for a 3rd year paper covering computing, artificial intelligence and pattern recognition, and it runs specialist modules in the 4th year on medical imaging, 3d computer graphics, statistical pattern processing, speech processing, computer vision and robotics.