IT-Porto
Towards an Applied Quantum Machine Learning
in Intelligent Transportation Systems
Networked Systems
Modeling and predicting (citywide) traffic congestion can have high
computational cost due to its complexity. Quantum machine learning (QML)
can aid speedup traffic dynamics by performing quantum operations. We aim
to implement a traffic flow application that would not be efficiently solved in a
classical computer. The main contributions are summarized as follows: (a)
data fusion method to infer microscopic fundamental diagrams which
allows us to determine the traffic state in individual road segments, (b)
methodology to identify traffic congestion causalities making use of Bayesian
networks to learn probabilistic dependencies from traffic data and
confounding factors, and (c) assess the performance of the proposed
methodology in a case study scenario in the city of Porto.
Can we implement Markov
computations faster in a
quantum simulator?
We explore Pauli X gate (NOT
gate) and Hadamard gate to
map the qubit-basis states |0⟩and |1⟩ to superposition states.
● Poster presented at the 3rd International Conference for Young Quantum Information Scientists 3-6 October 2017, Max Planck Institute
for the Science of Light, Friedrich-Alexander Universitat Erlangen-Nurnberg.
● Journal article at ITNOW (Oxford Academic)
● Extended abstract presented at Quantum Machine Learning Workshop at KDD 2018 - London, United Kingdom. 19 - 23 August 2018.
arXiv:1808.08429
● Conference article at 1st Workshop on Data-driven Intelligent Transportation (DIT 2018), Held in conjunction with IEEE ICDM 2018,
November 17-20, 2018 in Singapore, November 2018
Can we use quantum tabu search efficiently to find coupling maps
for quantum circuits?
We uncover optimal solutions and accelerate the optimisation process
using entangled states in a quantum optimisation.
UID/EEA/50008/2013
Six spatial windows for fusing data. Speed measurements for the virtual spatial windows. Probe Penetration Ratio (PPR): 2.92%.
Estimated microscopic fundamental diagrams for selected urban roads in the city of Porto.
Tree Augmented Naive
Bayes: congested (left),
uncongested (right) states
for the different locations.
Background and challenges
Simulations in Google Quantum Computing Playground and IBM Q Experience
Achievements
Traffic Congestion Using Fundamental Diagrams and Probabilistic Graphical Models
Description and main innovation