Differences

This shows you the differences between two versions of the page.

--- 2022-23:summer [2023/04/27 19:00] – ruppert
+++ 2022-23:summer [2023/04/28 22:46] (current) – ruppert
@@ Line 128: / Line 128: @@
 Optional: e-portfolio that demo previous projects that one has worked on
-==== How to use ChatGT in Classrooms ====
+==== How to use ChatGPT in Classrooms ====
 **Course:** EECS4080
@@ Line 221: / Line 221: @@
+==== Electric Load Forecasting via Deep Generative Models ====
+**Course:**  EECS4080
+**Supervisor:**  Michael Jenkin
+**Supervisor's email address:** jenkin@yorku.ca
+**Project Description:**
+With the fast increase in renewable energy generation and electric vehicles, electric load forecasting is becoming more and more important for power system operation. Based on the forecasting horizon, there are mainly three types of load forecasting, i.e., short-term, medium-term, and long-term. Short-term load forecasting mainly aims to predict the electric load in the next few seconds to the next few hours, which can be very helpful for real-world energy dispatching. In recent years, machine learning, especially deep learning, has shown impressive performance for short-term load forecasting.
+Generative models, e.g., generative adversarial networks, have shown great potential for computer vision and natural language processing. The potential of such generative models has not been well studied for load forecasting. In this project, we mainly aim to benchmark the performance of different types of deep generative models for short-term load forecasting. We will mainly work on OPEN EI data sets which consist of electric load consumption data sets for different buildings in the US.
+**Required skills or prerequisites:**
+Good python software skills. Interest in AI systems.
+**Recommended skills or prerequisites:**
+ Interest in GANs. Interest in AI software development.
+**Instructions:**
+Send CV, (unofficial transcript), GitHub repo address if available to Prof. Jenkin.
+==== Analyze the Impacts of Ensemble Learning for Anomaly Detection ====
+**Course:** EECS4080
+**Supervisor:**  Michael Jenkin
+**Supervisor's email address:** jenkin@yorku.ca
+**Project Description:**
+Hacking and false data injection from adversaries threaten can cause significant financial loss. Accurate detection of anomalies is of significant importance for the safe and efficient operation of modern power grids. In recent years, different types of techniques, such as statistical methods, unsupervised learning methods, generative models, and prediction-based methods, have been applied for anomaly detection. However, most of the current works assume the stability of the data distribution and ignore the distribution drift, which often happens in the real world.
+In this work, we aim to utilize the benefits of ensemble learning to address real-world anomaly detection problems. Specifically, we plan to dynamically utilize the different base models via ensemble learning to tackle the challenges of distribution drift in the real world. For this project, we will mainly work on two data sets, i.e., the Secure Water Treatment (SWaT) Dataset and ICS Cyber Attack Dataset.  These two data sets are frequently used real-world data sets for anomaly detection.
+**Required skills or prerequisites:**
+ Interest in AI systems. Interest in AI software development
+**Recommended skills or prerequisites:**
+Good python programming skills.
+Some course(s) in AI systems
+**Instructions:**
+Send CV, (unofficial transcript), GitHub repo address if available to Prof. Jenkin.