Skip to content

Commit eaba77b

Browse files
committed
1 parent 70c276f commit eaba77b

File tree

7 files changed

+41
-27
lines changed

7 files changed

+41
-27
lines changed

_config.yml

+1-1
Original file line numberDiff line numberDiff line change
@@ -24,7 +24,7 @@ author:
2424
avatar : "profile.png"
2525
name : "Zhengyang Jin"
2626
pronouns : # example: "he/his"
27-
bio : "AI Researcher (In progress)"
27+
bio : "AI Researcher (Beta)"
2828
bio1 : "BE STILL AND KNOW "
2929
location : "Singapore"
3030
employer : "A*STAR I2R"

_pages/about.md

+20-17
Original file line numberDiff line numberDiff line change
@@ -7,26 +7,28 @@ redirect_from:
77
- /about.html
88
---
99

10-
I am current a research engineer at A*STAR I2R Singapore. My research interests focus on Neuroscience-inspired AI, Adaptive Systems, and computational Neuroscience. My long-term goal is to abstract concepts such as consciousness, memory, and learning from biological neural systems and model them into advanced artificial intelligence models for application in real life. In turn, I also aim to use latest-generation AI technologies to assist and accelerate the exploration of biological neural systems, especially the brain, to enable fundamental science to explain the mechanisms of complex behaviors and systems. More specifically, my current research focuses on the Predictive Coding Network, a promising area where AI models utilize hierarchical learning processes akin to human cognitive functions. There are many novel structures, such as Active Predictive Coding and Hybrid Predictive Coding. I am now keenly interested in exploring and expanding the application of PCN in various scenarios, aiming to achieve state-of-the-art performance through structural adjustments or the integration of new modules.
10+
I am current a research engineer at A*STAR I2R Singapore. My research interests focus on Neuroscience-inspired AI, Adaptive Systems, and computational Neuroscience. My long-term goal is to abstract concepts such as consciousness, memory, and learning from biological neural systems and model them into advanced artificial intelligence models for application in real life. In turn, I also aim to use latest-generation AI technologies to assist and accelerate the exploration of biological neural systems, especially the brain, to enable fundamental science to explain the mechanisms of complex behaviors and systems.
11+
12+
My current research focuses on the Predictive Coding Network, a promising area where AI models utilize hierarchical learning processes akin to human cognitive functions. There are many novel structures, such as Active Predictive Coding and Hybrid Predictive Coding. I am now keenly interested in exploring and expanding the application of PCN in various scenarios, aiming to achieve state-of-the-art performance through structural adjustments or the integration of new modules.
1113

1214

1315
I am currently searching for appropriate PhD programs worldwide, aiming to fulfill part of my grand ambitions through them. I completed both my MSc and BSc in Artificial Intelligence at the University of Sussex, graduating with Distinction and First-class honors, respectively. Additionally, I undertook cross-disciplinary modules that provided me with unique insights into neuroscience, brain and behavior, artificial life, and autonomous vehicles, among others. These academic experiences have given me a solid foundation in AI and computational methods in computer science. I have also completed an internship at an AI startup company, hyperTunnel, which has endowed me with practical knowledge in multi-agent optimization, comprehensive analysis, solution architecture, and programming. These experiences demonstrate that I possess the academic and professional skills necessary to pursue a research position, and I look forward to fulfilling my aspiration for further academic development. Please contact me if you would like to discuss further with me anything about Artificial Intelligence, Neuroscience or robotics.
1416

15-
What I have Learnt:
17+
What I have Learned:
1618
------
1719

18-
| Artificial Intelligence | Data Science and Algorithms | Engineering | Programming Languages |
19-
| -------- | -------- | -------- | -------- |
20-
| Neural Networks | Multi Adaptive Regression Splines | Autonomous Vehicles | Python/Torch/Tensorflow|
21-
| Deep Learning | LOESS/LOWESS | Adaptive Control Systems | Matlab/Simulink |
22-
| Global Optimization | Gradient Boosting Machines | Image Processing | Html/Css/Javascript |
23-
| Computer Vision | XGBoost, LightGBM, CatBoost | Industrial Automation | C/C++/C#/.Net/Unity3D |
24-
| Reinforcement Learning | Bayesian Networks | Robotics optimization | Git/Docker/Bash/Shell |
25-
| Language Models | Ensemble Methods(Bagging...) | Swarm Intelligence | Java/SpringBoot |
26-
| Bio-inspired AI | Variational Autoencoders | | |
27-
| Reasoning and Knowledge | Causal Impacts | | |
28-
| Reinforcement Learning | Graph Neural Networks | | |
29-
| Free Energy Methods | Evolutionary Algorithms | | |
20+
| Artificial Intelligence | Data Science and Algorithms | Engineering |
21+
| -------- | -------- | -------- |
22+
| Neural Networks | Multi Adaptive Regression Splines | Autonomous Vehicles |
23+
| Deep Learning | LOESS/LOWESS | Adaptive Control Systems |
24+
| Global Optimization | Gradient Boosting Machines | Image Processing |
25+
| Computer Vision | XGBoost, LightGBM, CatBoost | Industrial Automation |
26+
| Reinforcement Learning | Bayesian Networks | Robotics optimization |
27+
| Language Models | Ensemble Methods(Bagging...) | Swarm Intelligence |
28+
| Bio-inspired AI | Variational Autoencoders | |
29+
| Reasoning and Knowledge | Causal Impacts | |
30+
| Reinforcement Learning | Graph Neural Networks | |
31+
| Free Energy Methods | Evolutionary Algorithms | |
3032

3133
MISCELLANEOUS
3234
------
@@ -36,11 +38,12 @@ MISCELLANEOUS
3638
| Python/Torch/Tensorflow| Baking | Labs |
3739
| Matlab/Simulink | Cocktail Making | Bars |
3840
| Html/Css/Javascript | Tennis | Gardens |
39-
| C/C++/C#/.Net/Unity3D | Swimming | tennis courts |
40-
| Git/Docker/Bash/Shell | Cello | Gaming Room |
41-
| Java/SpringBoot | Gaming | Practice Room |
41+
| C/C++/C#/.Net/Unity3D | Swimming | Seaside |
42+
| Git/Docker/Bash/Shell | Cello | Kitchen |
43+
| Java/SpringBoot | Gaming | |
4244

4345
For more info
4446
------
47+
4548
More info about some of my past projects or research experiences? You can find them at [Shawcase](https://dashpulsar.github.io/portfolio/).
4649
Want to get in touch with me? You can find the way you want at [Contact Me](https://dashpulsar.github.io/teacher/).

_pages/cv.md

+5-6
Original file line numberDiff line numberDiff line change
@@ -11,19 +11,18 @@ redirect_from:
1111

1212
Education
1313
======
14-
* University of Sussex, 2021-2022 <br>&nbsp;
14+
* **University of Sussex**, 2021-2022 <br>&nbsp;
1515
MSc in Artificial Intelligence and Adaptive Systems <br>&nbsp;
1616
Grade: Distinction, GPA: 3.7/4.0
17-
* University of Sussex, 2018-2021 <br> &nbsp; BSc in Computer Science and Artificial Intelligence <br>&nbsp; Grade: First-class, GPA: 3.7/4.0
17+
* **University of Sussex**, 2018-2021 <br> &nbsp; BSc in Computer Science and Artificial Intelligence <br>&nbsp; Grade: First-class, GPA: 3.7/4.0
1818

1919
Work Experience
2020
======
21-
* HyperTunnel, Mar 2023 - Aug 2023, AI Engineer Intern.
21+
* **HyperTunnel**, Mar 2023 - Aug 2023, AI Engineer Intern.
2222
* Developed and optimized swarm intelligence evolutionary algorithms, enhancing the efficiency of surveying robots for underground construction. Dynamic adaptive optimization of multi-objective functions.
2323
* Implementation of AI-driven solutions for dynamic excavation tasks, improving geological detection and operational precision.
24-
* Supervisor: Xinghui Tao
2524

26-
* Microsoft, Oct 2021 - Jun 2022, Mentee of EMBRACE Mentoring Program
25+
* **Microsoft**, Oct 2021 - Jun 2022, Mentee of EMBRACE Mentoring Program
2726
* Applying fairness ML algorithms to improve the feedback of products or advertisements among diverse populations.
2827
* Explore users' preferences and habits with reinforcement learning, adjusting the product's features and interface, to achieve a better user experience.
2928

@@ -40,7 +39,7 @@ Research Experience
4039
Skills
4140
======
4241
* Programming Languages
43-
* Proficient in Python, MATLAB, Java, and R
42+
* Proficient in Python, MATLAB, PyTorch
4443
* Experienced with C, C++, C#, JavaScript, Unity3D
4544
* Software and Data Management
4645
* Git, Docker, SQL, Linux

_portfolio/portfolio-1.md

+15-3
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,19 @@
11
---
2-
title: "Comparing Gillespie and Meanfield Simulation"
3-
excerpt: "Short description of portfolio item number 1<br/><img src='/images/500x300.png'>"
2+
title: "Evolution controller for foraging behaviour"
3+
excerpt: "Optimize the foraging behavior of robotic controllers through reinforcement learning and evolutionary algorithms.<br/><img src='/images/P1.png'>"
44
collection: portfolio
55
---
66

7-
This is an item in your portfolio. It can be have images or nice text. If you name the file .md, it will be parsed as markdown. If you name the file .html, it will be parsed as HTML.
7+
This project investigates the development of a robot capable of adapting to a dynamic and challenging simulated environment using principles derived from evolutionary algorithms. The robot, inspired by Braitenberg's vehicles, is designed to navigate a simulated world where it must avoid traps and seek resources like food and water, which can unpredictably turn into deadly poisons. This setup tests the robot's ability to continuously adapt to changing conditions.
8+
9+
![P12](https://dashpulsar.github.io/images/P12.png)
10+
11+
The method involves creating a robot with simple sensorimotor links, equipped with two batteries and sensors that respond to different simulated "light sources" representing food, water, and poison. The robot's actions are governed by the outputs of these sensors processed through a genetically encoded activation function. This function evolves across generations to improve the robot’s performance in foraging and avoiding hazards.
12+
13+
The robot operates on a 10x10 map where the location of resources and hazards is randomized. Its survival hinges on efficiently gathering food and water while dodging poisons. The evolutionary process optimizes the sensor activation functions using a genetic algorithm that considers the robot's battery life and foraging efficiency.
14+
15+
![P13](https://dashpulsar.github.io/images/P13.png)
16+
17+
Results show that the evolved robot can adeptly navigate its environment, making strategic decisions based on resource availability and battery status. It exhibits behaviors like preference for closer and safer resources and a shift in resource gathering strategy based on battery levels.
18+
19+
The discussion highlights the robot's proficiency in foraging and hazard avoidance and suggests potential future enhancements such as dynamic control systems and competitive scenarios with multiple robots. This project illustrates the potential of evolutionary algorithms in developing autonomous robots that can adapt to complex and unpredictable environments.

images/P11.png

32.6 KB
Loading

images/P12.png

24.2 KB
Loading

images/p13.png

71.7 KB
Loading

0 commit comments

Comments
 (0)