Dato’ Dr. Haji Amirudin Abdul Wahab (“Dato’ Dr. Amir”) – Human and Visual Communication: The Brief History of Data Visualisation
Visualizations have been around in human history. It delivers a story about human life through data. The great visualization provides communication narrative through connections among data points, reframing the story about data – often more than words could describe the story – so that human could see the important insights in many perspectives.
What makes the visualization is great? It’s about human perception towards filling missing information in our brain. It’s about human eye perceives visual elements – to extract new information from aggregated data points.
The intellectual history of data visualization from medieval to modern times provide us with great lessons on how human use visualization as communication tool to tell a story.
Prof. Dr. Sergio A. Velastin – Vehicle Detection using Alex Net and Faster R-CNN Deep Learning Models: A Comparative Study
This talk presents a comparative study of two deep learning models used here for vehicle detection. AlexNet and Faster R-CNN are compared with the analysis of an urban video sequence. Several tests were carried to evaluate the quality of detections, failure rates and times employed to complete the detection task. The results allow to obtain important conclusions regarding the architectures and strategies used for implementing such network for the task of video detection, encouraging future research in this topic.
Prof Dr Patrice Boursier – Mitigating and Trying to Prevent Natural Disasters Using Social Media, Big Data, IoT and Artificial Intelligence
Natural disasters are becoming more and more frequent and costly. Flash floods and coastal floods are two of the most disastrous events that may occur in many different regions of the world. A large and increasing fraction of the world’s population lives along coastlines, and significant numbers are already vulnerable to coastal flooding in both the developed and developing world. This exposure is likely to increase as sea level rises with climate change. In some locations, future changes in tropical storms and storm surges may further add to the risk of flooding. Predicting floods might help taking necessary precautions and save human lives.
New technologies, namely Big Data, Internet of Things and Machine Learning make it now easier to deal with natural disasters. Geographic Information Systems (GIS) have also been used for more than 30 years for managing the territory with geo-referenced or spatial data, and they can be combined with “big” data management systems and mathematical models.
Preparation and mitigation benefit from these tools, but what about prediction and prevention? How far can we go with these new technologies? The objective of this talk will be to review the progress made possible during the last few years, and what we can expect in the near future.
Dr Tomas Maul – To compose or decompose? Deep and broad learning via parallel-circuit neural network
The aim of this talk is to introduce the audience to parallel-circuit artificial neural networks (PCANNs). At the start, the visual system background, and the natural computational motivation of PCANNs will be laid out, followed by a detailed explanation of the architecture and why it is needed. Mechanisms such as dropout and dropcircuit, which assist parallel-circuit networks in boosting generalization performance as compared to standard single-circuit networks, will also be discussed. Several experimental results will be summarized and implications with regards to node composition and decomposition (and deep and broad learning) will be provided as a conclusion.