Zhang Zhang

PhD student. AI Machine learning Engineer/ Research Associate

Stockon-on-tees, United Kingdom

Summary

I am currently a final year PhD student focus on Research of Machine Learning at Teesside University. I am seeking a position as Machine Learning Research Engineer.
• My current research area(PhD thesis) is the strategy of collective intention recognition of multi-agent system in machine intelligence. The current improved key algorithms during my PhD study follow as,
– Build Multi-agent Model: Integrating the Markov decision process(MDP) model into the Belief-Desire-Intention(BDI) agent framework to build up a multi-agent decision model.
– Intention Recognition Process: Constructing the Intention Classifier for the multi-agent intention recognition process.
– Collective Intention Recognition Strategy: Building the Collective Behaviour Pattern for recognising the collective intention type based on the intention recognition process of the group agents. Combining with the MDP model to help the agents to choose the optimal cooperative plans for their current situation.

Languages:

Chinese, English

Favorite Python Packages:

numpy, sklearn, pandas, hashlib, matplotlib, cPickle,scipy, scikit-learn, bokeh,jupyter notebook

Experience

Research Experience

• Designing Planning Models for the interactions of Multi-agent System

Research Associate/part-time, University of Teesside 04/2019 - 07/2019

– Designing a planning model for improving agents’ automation in their interactions based on the interactive dynamic influence diagrams(I-DIDs) and Partial Observation Markov Decision Process(POMDP) model. Through constructing the policy tree to learn behaviours of agents. Using the partial policies to match the behaviour equivalence to predict the interactions of agents.

• Collective Intention Recognition Strategy for Multi-agent System

PhD Student/full-time, University of Teesside 10/2015 - 02/2020

– In order to solve the collective intention recognition among multiple agents and the cooperative decision-making strategy of groups, building the collective behaviour pattern based on cooperative relationships as a template to identify whether multiple agents have cooperative intention or not. Improved fast search and pattern match algorithm are used to assist multiple agents quickly find appropriate partners to form collective teams. Moreover, MDP model is used as a controller to choose the optimal cooperative strategy for collective teams according to their current situation.

• Intention Recognition Process for Multi-agent System

University of Teesside 10/2015 - 02/2020

– Constructing the intention classifier with the extracted action feature sub-trees as the judgement attributes based on the decision tree algorithm for solving the efficiency problem of simultaneous intention recognition process of multi-agent system. Building a officebuilding game demo to compare the efficiency of traditional Bayesian Network Approach(BNS) and intention classifier in recognising multi-agent intentions and grouping them. In addition, in this simulation, the observer will compare the differences in efficiency and accuracy of recognising the intention of individuals or group agents

• Building Multi-agent Decision Model

University of Teesside 10/2015 - 02/2020

– In order to solve the global decision-making problem of multiple agents, MDP model is integrated into BDI framework to build a multi-agent decision-making model. And in this process, analyse the advantages of the integration of these two frameworks. In addition, building a tileworld game demos to analyse the influence on the change of environment factor to the results of intention reconsideration process of multiple agents who have different attributes according to the change of environmental factors.

Working Experience

• Android application developer/internship Beijing, China BeiJing Sixth Electronics Research Institute 06/2013 - 06/2014

– Develop the Android application and Communicate with customers to discuss the plan of

project and solve the problems according to the requirements from customers. Through my

self-management to organise my time and learn much knowledge and technology from team

work. For example, photoshop, HTML5, Android.

• Java Software Developer/part-time Beijing, China BeiJing TangMuYiTong Technology Company 08/2010 - 08/2013

– Develop Java web project and Communicate with customers to discuss the plan of project and solve the problems according to the requirements from customers. Plan and organise the part of project according to the requirements of project. For example, the frame of the application, UI design. Through my self-management to organize my time and learn much knowldge and technology from team work. For example, photoshop, HTML5, Java, JSP and SSH.

Skills

Apache, DevOps, Django, Docker, Git, Hadoop, Keras, Linux, Machine Learning, MySQL, NumPy, Pandas, SciPy, TensorFlow, Virtualenv