Mohammad Ashraful Anam
Research Associate
I obtained my BSc degree in Electrical and Electronic Engineering at the Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh in 2003 and the MSc degree in Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh in 2008. I joined the CISG (Communications and Information Systems Group) at the Department of Electronic and Electrical Engineering (EEEng) of the University College London (UCL) in 2010, where I completed my Ph.D. degree in 2014 in the field of throughput scaling of digital signal processing based on desired precision under the supervision of Dr. Yiannis Andreopoulos. I am currently working as a research associate in the same university.
In my research work, I am investigating novel ways for: (i) substantially accelerating computations (or reducing energy consumption) when pursuing approximate results; (ii) allowing for aggressive frequency or voltage scaling in hardware while ensuring that the obtained results are reliable.
Within the first part, I have derived for the first time a systematic approach to scale energy and processing cycles when computing approximate convolution realizations via state-of-the-art numerical libraries, such as Intel Integrated Performance Primitives (IPP). My approach demonstrated up to 3-fold improvement in processing throughput against the current state-of-the-art and, for this reason, it was published in the IEEE Transactions on Multimedia as a full paper. This work also led to a 2012 summer internship at ARM (Cambridge).
Within the second part, I have recently invented a new approach for highly-reliable linear, sesquilinear and bijective (LSB) processing of integer data streams that is based on the novel concept of numerical entanglement. This work led to a patent application, sponsored by UCLBusiness, UCL's wholly-owned subsidiary responsible for commercializing inventions of UCL researchers.
  • Failure Mitigation in Linear, Sesquilinear and Bijective Operations On Integer Data Streams Via Numerical Entanglement
    M.A. Anam, Y. Andreopoulos
    21st IEEE International On-Line Testing Symposium (IOLTS 2015), 6-8 July 2015
  • Method and Apparatus for the Detection of Faults in Data Computation
    M. A. Anam, Y. Andreopoulos
    Patent Application, Sponsor: UCLBusiness.
  • Precision--Energy--Throughput Scaling Of Generic Matrix Multiplication and Convolution Kernels Via Linear Projections
    M.A. Anam, P.N. Whatmough, Y. Andreopoulos
    IEEE Transactions on Circuits and Systems for Video Technology, pp. 1860-1873, vol. 24, no. 11, Nov. 2014
  • Precision-energy-throughput scaling of generic matrix multiplication and discrete convolution kernels via linear projections
    M.A. Anam, P.N. Whatmough, Y. Andreopoulos
    IEEE 11th Symposium on Embedded Systems for Real-time Multimedia (ESTIMedia) 2013, pp.21,30, 3-4 Oct. 2013
  • Highly-reliable integer matrix multiplication via numerical packing
    I. Anarado, M.A. Anam, D. Anastasia, F. Verdicchio, Y. Andreopoulos
    IEEE 19th International On-Line Testing Symposium (IOLTS) 2013, pp.19,24, 8-10 July 2013
  • Throughput Scaling Of Convolution For Error-Tolerant Multimedia Applications
    M. A. Anam, Y. Andreopoulos
    IEEE Transactions on Multimedia, vol.14, no.3, pp. 797-804, June 2012
  • Throughput/Precision Computation Of Convolution In Programmable Processors
    M. A. Anam, Y. Andreopoulos
    London Communications Symposium (LCS 2011), UCL, London
Department of Electronic and Electrical Engineering
Roberts Building
UCL (University College London)
Torrington Place
London
WC1E 7JE

Room: 804 (eight floor)

E-mail: uceeman (at) ee (dot) ucl (dot) ac (dot) uk

UK Telephone: 020 7679 2123
Internal Telephone: 32123
International Telephone: +44 20 7679 2123

UK Fax: 020 7388 9325
International Fax: +44 20 7388 9325