Ji Junkai, PhD
Assistant Professor, Shenzhen University
Email: jijunkai@szu.edu.cn
Education Background:
2013, Hefei University of Technology, Chemical Engineering and Technology, Bachelor of Engineering;
2016, University of Toyama, Japan, Intelligent Information Engineering, Master of Engineering;
2018, University of Toyama, Japan, Mathematical and Human Systems Science, PhD.
Research Interests:
Artificial intelligence drug design, brain-inspired neural networks;
Published over 70 journal and conference papers, applied for 4 patents;
Principal Investigator: National Natural Science Foundation of China (NSFC) General Program, Youth Program, Guangdong Provincial NSFC General Program, Joint Fund Youth Program, Shenzhen Natural Science Foundation General Program.
[1] Junkai Ji, Jin Zhou, Zhangfan Yang, Qiuzhen Lin, and Carlos A. Coello Coello. "AutoDock Koto: A Gradient Boosting Differential Evolution for Molecular Docking." IEEE Transactions on Evolutionary Computation(2022).
[2] Junkai Ji, Yajiao Tang, Lijia Ma, Jianqiang Li, Qiuzhen Lin, Zheng Tang, and Yuki Todo. "Accuracy Versus Simplification in an Approximate Logic Neural Model." IEEE Transactions on Neural Networks and Learning Systems(2020).
[3] Junkai Ji, Jiajun Zhao, Qiuzhen Lin, and Kay Chen Tan. "Competitive Decomposition-Based Multiobjective Architecture Search for the Dendritic Neural Model." IEEE Transactions on Cybernetics(2022).
[4] Junkai Ji, Minhui Dong, Qiuzhen Lin, and Kay Chen Tan. "Noninvasive Cuffless Blood Pressure Estimation With Dendritic Neural Regression." IEEE Transactions on Cybernetics (2022).
[5] Junkai Ji, Minhui Dong, Qiuzhen Lin, and Kay Chen Tan. "Forecasting Wind Speed Time Series Via Dendritic Neural Regression." IEEE Computational Intelligence Magazine 16, no. 3 (2021): 50-66.
[6] Junkai Ji, Shuangbao Song, Cheng Tang, Shangce Gao, Zheng Tang and Yuki Todo*, "An artificial bee colony algorithm search guided by scale-free networks." Information Sciences, 473 (2019): 142-165.
[7] Junkai Ji, Shuangbao Song, Yajiao Tang, Shangce Gao, Zheng Tang and Yuki Todo, "Approximate logic neuron model trained by states of matter search algorithm." Knowledge-Based Systems, 163 (2019): 120-130.
[8] Junkai Ji, Shangce Gao, Jiujun Cheng, Zheng Tang, and Yuki Todo, "An approximate logic neuron model with a dendritic structure." Neurocomputing173 (2016): 1775-1783.
[9] Junkai Ji, Cheng Tang, Jiajun Zhao, Zheng Tang, and Yuki Todo. "A Survey on Dendritic Neuron Model: Mechanisms, Algorithms and Practical Applications. Neurocomputing (2022).
Personal Homepage:
https://www.researchgate.net/profile/Junkai-Ji.
Recruiting Graduate and Doctoral Students:
Recruiting graduate students in Computer Science, Mathematics, Medicine, and Biology;
Training Objectives:
(1) Basic Knowledge: Mastering solid basic theories and systematic professional knowledge in this discipline, as well as systematic bioinformatics methods and experimental skills, and mastering one foreign language;
(2) Basic Academic Abilities: Publishing one academic paper related to the major as the first author or the first author other than the supervisor, or obtaining one invention patent that has passed substantive examination;
(3) Basic Requirements for Dissertations: The dissertation should demonstrate the author's preliminary mastery of scientific research methods and experimental techniques in this research direction, and the ability to independently conduct scientific research.
Postdoctoral Fellowship Recruitment:
We are recruiting postdoctoral fellows under the age of 35 who have obtained their doctoral degrees from high-level universities or research institutions at home and abroad within the past three years; those with research backgrounds in machine learning, drug design, etc., will be given priority.
(1) Salary and Benefits: Annual salary starting from RMB 320,000; Shenzhen provides a subsidy of RMB 300,000 to postdoctoral fellows upon completion of their postdoctoral research for research investment or early-stage business expenses.
(2) Career Development: Qualified postdoctoral fellows can apply for professional technical qualification assessment; those who perform exceptionally well during their postdoctoral research and meet the recruitment requirements may be selected for tenure-track faculty positions.
(3) Housing Policy: Postdoctoral fellows can apply for temporary housing provided by the university with discounted rent.
(4) Household Registration Policy: Postdoctoral fellows can register their household in Shenzhen; their spouses and minor children can also apply for household registration.
(5) Postdoctoral fellows can apply for research grants at various levels as project leaders during their postdoctoral research.
Interested applicants are welcome to contact us via email.
Research Directions
Computer-Aided Drug Design
Targeting disease-specific proteins, we utilize deep learning and machine learning methods to screen for effective candidate drug molecules, guiding subsequent pharmacological and clinical trials.
Ligands are drug molecules; dosages of a few micrograms can significantly alter bodily functions, such as aspirin. Proteins are carriers involved in human biochemical reactions, such as targets for certain cancers.
Autodock Koto is a molecular docking algorithm developed by our team based on gradient differential evolution. Compared with either commercial or academic docking programs, Koto yields dramatic improvements in the success rates of generating crystal-like complex conformations.
Brain-Inspired Neural Networks
Inspired by biological neurons and neural circuits, we model artificial neural network architectures with biological interpretability.
The dendritic neuron model is a novel neural model we propose, featuring a plastic dendritic morphology.
The dendritic neuron model consists of synaptic layers, dendritic layers, membrane layers, and cell bodies. Each layer performs corresponding neural functions through different activation functions.
Through neural pruning schemes, DNM can eliminate redundant synapses and dendritic branches to simplify its architecture and form unique neuronal morphologies for each specific task;
Through logical approximation schemes, DNM can be transformed into a logic circuit classifier (LCC), which consists only of comparators and logical AND, OR, and NOT gates;
LCC is easily implemented in hardware for massively parallel computation, such as field-programmable gate arrays (FPGAs) and very large-scale integration (VLSI);
For model details, related papers, and source code of DNM, please see the model website: https://jijunkai123.github.io/DNM/.
