International Symposium on Artificial Intelligence and Machine Learning

9th October, 2017

The “International Symposium on Artificial Intelligence & Machine Learning” was held at the NUST School of Electrical Engineering & Computer Science on Monday October 09, 2017. Three visiting researchers from Germany gave talks on topics related to Artificial Intelligence & Machine Learning at this symposium. The symposium received an overwhelming response from SEECS students with over 250 students in attendance throughout the workshop. The symposium speakers were impressed by the high turnout of students and thoroughly enjoyed interacting with SEECS students. This report summarizes the proceedings of the symposium.

Prof. Dr. Marcus Liwicki, currently a professor at Technical University of Kaiserslautern and a senior assistant at University of Fribourg, Switzerland, gave a talk on “Automatic Handwriting Recognition, With & Without Deep Learning”. He started off with an intuitively understandable, but fairly in-depth explanation of what Machine Learning essentially means. Progressing from the explanation of how neural networks mimic the neurons in the human brain, Dr. Liwicki went on explaining the utility of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) in Optical Character Recognition (OCR); which is used for conversion of digitized printed text (e.g., scanned documents or images captured via a camera) into editable and searchable text. He explained how building on top of the models used for printed text recognition, we have advanced to very accurately recognizing handwritten text, particularly that retrieved from historical Latin script. He talked in detail about binarization and background estimation techniques, and how these help enhancing the overall performance of handwriting recognition systems. Dr. Liwicki also conducted an additional 60 minute (hands-on) session on deep learning where he deeply described the theory of CNN and how they can be used for various practical tasks. He kindly made his slides and references, as well as course material, available under:

Prof. Dr. Ulrich Schwanecke, Professor for Computer Graphics and Vision at the RheinMain University of Applied Sciences Wiesbaden Rüsselsheim, gave an engaging talk on “Model-based 3D object tracking”. His talk was mostly centered around 3D Computer Vision related techniques. He explained the significance of Euclidean and projective geometry, different coordinate systems, and how the spatial position and orientation (the Pose) of an object can be described. Building on those concepts, he explained the naïve point based methods as well as the state-of-the-art methods based on scene segmentation using histogram information and truncated signed distance functions of the boundary of the silhouette of the object, used for determine the Pose of an object. He demonstrated that the presented techniques can be used for object augmentation in live video feeds. He went on to mention real-time techniques used to cater for object occlusion in the augmentation process.

Prof. Dr. Adrian Ulges is a Professor at RheinMain University of Applied Sciences (HSRM), with interests in applied mathematics and machine learning. He delivered a talk on “Multi-modal photo-graphics retrieval”centered on applications of deep learning to content-based image retrieval. He also described introductory basics to deep convolutional neural networks (DeepCNNs), and pointed out application aspects on how to use pre-stored 3D models to find furniture objects in natural images as a use case. Applications of adversarial models in this domain were also discussed.

Dr. Hassan Aqeel Khan, an Assistant Professor at NUST SEECS, conducted a session on the basics of Machine Learning and its applications in Digital Pathology. Attendees were introduced to the basic tasks performed for landscape quantification e.g. Nucleus segmentation, Mitosis detection and their significance in diagnosis, prognosis and grading of cancers. Challenges and opportunities in this domain were also discussed and the audience was given pointers on various, useful, cancer image repositories.

Dr. Muhammad Ali Tahir who is an Assistant Professor at NUST SEECS conducted a session on automatic speech recognition (ASR). He introduced the basics of ASR. This was followed up by in depth discussion on interesting topics such as Hidden Markov Models, Deep Neural Networks and their applications in Automatic Speech Recognition.

Dr. Faisal Rashid, who is an Assistant Professor at UET Lahore, introduced Internet of Things (IoT) and applications of Machine Learning in this domain. He discussed a number of interesting applications of IoT and talked in detail about how Machine Learning could be used to derive maximum benefit from IoT devices in the near future.

Wrap-up Meeting

A final meeting was held between Principal SEECS (Dr. S. M. Hassan Zaidi), SEECS researchers and the visiting German Professors about the potential of future collaborations between SEECS and the research groups of the visiting Professors. Listed below are some of the key outcomes of this meeting:

  1. Dr. Liwicki demonstrated a lot of interest in research related to STEAM (Science, Technology , the Arts and Mathematics) teaching. Principal SEECS highlighted that Dr. Mudassir Malik of the DoC department at SEECS is actively conducting research in this domain. It was suggested that the Dr. Mudassir be connected with Dr. Liwicki soon to identify mutual areas of interest.
  2. Dr. Hassan Aqeel will have a more detailed discussion with Dr. Liwicki to set up collaboration in the medical imaging domain.
  3. Dr. Imran Malik will explore establishment of collaboration in Digital Forensics.
  4. Dr. Faisal Shafait will continue to strengthen the existing collaboration with Dr. Adrian Ulges and Ulrich Schwanecke in the area of Cross Modal Information Retrieval.

At the end of the meeting, the visiting delegates were bid farewell by Principal SEECS and proceeded to dinner with SEECS faculty members who organized the event to discuss mutual areas of interest in an informal setting. The German guests thank NUST for the kind hospitality.