Recordings of this Conference are now available with closed captioning on our Youtube Channel: https://www.youtube.com/user/IMAmaths/videos
10:00-11:00 – Prof Mini C. Saaj – Robots and Maths: Are they related?
11:00-12:00 – Prof Danilo Mandic – Tensors, graphs, and deep networks: Convergence of concepts and ideas
For a number of years, the IMA has been running a series of conferences to promote mathematics with the aim of demonstrating to both mathematicians and non-mathematicians the many uses of modern mathematics.
This summer, due to Covid19, IMA Mathematics 2020 will be run virtually in a series of weekly two hour events, starting on Tuesday 14 July 2020 with the IMA Presidential Address. Other talks will cover areas such as defence, regulation, privacy enhancing technologies and modelling in industrial maintenance and reliability.
Recordings of this Conference are now available with closed captioning on our Youtube Channel: https://www.youtube.com/user/IMAmaths/videos
Abstracts
Prof Mini C. Saaj – Robots and Maths: Are they related?
The philosophy of mathematics drives the art of designing an intelligent robot. However, engineering a robot is a laborious exercise, but understanding its underlying principle can be made easy by a mathematician. In this talk, you will hear about the deeper bonding between robotics and engineering mathematics. You will also gain a more in‐depth insight into how software programming helps with translating mathematical models into real‐life robots in action
Speaker Biography
Prof Mini C. Saaj is the Global Chair in Robotic Engineering at the University of Lincoln, UK, where she is also the lead for research in Industrial Digitalisation and System Intelligence. Previously, she was the Head of Robotics and Control research group and the Director of Post Graduate Research at the Surrey Space Centre, UK. Her expertise is in modeling and control of spacecraft, design and control of rigid and flexible manipulators, systems engineering using Model‐Based Systems Engineering approach, design and control of wheeled and legged planetary rovers, bio‐robotics and flexible medical robotics. Prof. Saaj has secured research grants over £4M as lead and co‐ investigator. In addition to publishing 102 articles, she has successfully supervised 20 post‐doctoral Research Fellows and PhD students.
Being a leading female Space engineer and Roboticist, Prof. Saaj actively promotes Space Engineering education. She was a Flying Lecturer with the EngineeringUK for the ‘Engineers make it happen’ campaign (2008‐2010). She won the Airbus ‐ Royal Academy of Engineering Secondment award (2009) and the University of Surrey Vice Chancellor’s Teaching Excellence award in 2013. Prof. Saaj is a Chartered Engineer with the Engineering Council, UK and a Senior Member of IEEE and AIAA and a member of IEEE Women in Engineering. More details can be found at:
https://staff.lincoln.ac.uk/f3bf7246‐9b69‐4e6c‐9157‐f8c4d7221fc3
www.linkedin.com/in/MiniCSaaj
https://www.surrey.ac.uk/people/chakravarthini‐mini‐saaj
Prof Danilo Mandic – Tensors, graphs, and deep networks: Convergence of concepts and ideas
The widespread use of multisensor technology and the emergence of big data sets have highlighted limitations of standard flat-view matrix models in terms of their expressiveness, and the necessity to move toward more versatile data analysis tools. This has also highlighted the limitations of the rigid nature of standard uniform sampling in time (for time series) and space (for images and sensor networks), both prohibitive to efficient modelling of data recorded on irregular domains. These, together with the computational bottleneck associated with the deep learning paradigm, are calling for a unifying framework for the analysis of big data acquired on non-uniform domains and at an affordable computational costs. This talk addresses the convergence of concepts of tensor decompositions and data analytics on graphs, and offers insights into how their joint treatment provides feasible means to mitigate the curse of dimensionality associated with both the big data and deep learning paradigms. We show that for data which exhibit underlying signal generation structure, this can lead to orders of magnitude savings in the storage and computational complexity (super-compression), together with natural data separability by virtue of multilinear tensor algebra. It is further shown how the concept of local neighbourhood, an intrinsic feature of data acquired on graphs, can be employed to both perform dimensionality reduction and perform cost-efficient clustering directly on the domains where data reside. Finally, as an example, the convergence of these two concepts is shown to both introduce physical meaning into the otherwise black-box nature of neural networks, and to offer enhanced interpretability throughout the processing chain and at an affordable computational cost.
Speaker Biography
Danilo P. Mandic is a Professor with Imperial College London, UK, and has been working in the areas of Statistical Signal Processing and Machine Intelligence. He is a Fellow of the IEEE and hs serevd on the Board of Governors of the International Neural Networks Society (INNS). Prof Mandic is a Co-Director of the Financial Signal Processing Lab at Imperial, has more than 500 publications in journals and conferences, and has received President’s Award for excellence in postgraduate supervision at Imperial. His work related to this talk includes his book “Recurrent Neural Networks for Prediction”, Wiley 2001, two recent monographs on Tensor Networks for Dimensionality Reduction and Large Scale Optimisation (Now Publishers, 2016 and 2017), and ongoing work on Data Analytics on Graphs https://arxiv.org/abs/1907.03467. Prof Mandic is the 2019 recipient of the Dennis Gabor Award, given by the International Neural Networks Society (INNS) for Outstanding Achievements in Neural Engineering, and a recipient of the 2018 Best Paper Award in IEEE Signal Processing Magazine for the article “Tensor decompositions for signal processing applications: From two-way to multiway component analysis.”
Registration
Registration is currently open at https://my.ima.org.uk/
If you are an IMA Member or you have previously registered for an IMA conference, then you are already on our database. Please “request a new password” using the email address previously used, to log in.
Support
The IMA Mathematics 2020 Online Series has been organised in collaboration with the Newton Gateway to Mathematics.
Further information
For general conference queries please contact the Conferences Team
E-mail: conferences@ima.org.uk
Tel: +44 (0) 1702 354 020
Institute of Mathematics and its Applications, Catherine Richards House, 16 Nelson Street, Southend-on-Sea, Essex, SS1 1EF, UK.
