Projects in Quantum Engineering

Artificial Intelligence (AI) systems, particularly those based on deep neural networks (DNNs), can be interpreted as nonlinear function approximators. The expressive power of AI in engineering applications largely stems from its capacity to handle complex nonlinear relationships. In current DNN architectures, this nonlinearity is primarily introduced through fixed activation functions such as ReLU, Sigmoid, Tanh, GELU, Swish, and Leaky ReLU. While these functions are simple and efficient, they offer limited flexibility and nonlinear expressiveness. As a result, large-scale DNNs with many layers and parameters are often required to model intricate nonlinear phenomena, leading to increased computational cost and training complexity, and often without significantly improving approximation quality. 

To address these limitations, we have recently developed an analytic, learnable activation function capable of approximating any continuous nonlinear function to arbitrary precision. This development opens the door to a new class of compact, high-capacity neural networks with greatly enhanced nonlinear modeling power. When integrated with mathematical frameworks such as the Kolmogorov Representation Theorem, this novel activation mechanism enables the design of small-scale yet powerful DNNs that can approximate complex engineering functions that are otherwise intractable or require prohibitively large models. 

Leveraging our learnable activation functions, this project aims to develop a new AI framework tailored for engineering applications, where more efficient, compact, and theoretically grounded AI systems can be built for engineering design, control, and optimization. 

Supervisor: Prof. En-Hui Yang 
Email: ehyang@uwaterloo.ca 

This project aims to build deep machine learning models (e.g. recurrent neural networks, transformer models, etc.) to perform experimental data analysis where tasks of classification and pattern recognition are performed. 

Required knowledge & skills: Python, machine learning algorithms.

Supervisor: Prof. Na Young Kim
Email: nayoung.kim@uwaterloo.ca
Location: Remote/RAC-2101

This project aims to build quantum and classical deep machine learning models (e.g., recurrent neural networks, transformer models, etc.) to assess real-life big data, where tasks such as classification, anomaly detection, and pattern recognition are performed. 

Required skills: Python, machine learning algorithms.

Supervisor: Prof. Na Young Kim
Email: nayoung.kim@uwaterloo.ca
Phone: 519-888-4567 x30481
Location: Remote/RAC-2101

This project aims to apply and adapt existing denoising algorithms to scientific datasets, such as medical or chemistry data. The goal is to improve data quality and prepare it for subsequent analysis in larger research pipelines. Students will start by using existing Python code for denoising (e.g., for 1D signals or 2D images) and are encouraged to explore alternative preprocessing methods to achieve higher-quality results. This project provides hands-on experience in data handling, preprocessing, and experimental validation within real scientific workflows.

Required knowledge & skills: Basic Python programming; script automation; familiarity with data processing or machine learning libraries; interest in scientific or medical data analysis.

Supervisor: Prof. Na Young Kim

Email: nayoung.kim@uwaterloo.ca

Phone: 519-888-4567 x30481

Location: Remote/RAC-2101

It is hypothesized that the gradient map of a benign image will be very different from that of a corresponding adversarial example in terms of their statistic properties, spatial shapes, and visualization. This project will first analyze the differences between the gradient maps of benign images and the gradient maps of corresponding adversarial examples. Based on the analysis, the project will then explore novel ways to detect adversarial examples through gradients. 

Supervisor: Prof. En-Hui Yang 
Email: ehyang@uwaterloo.ca 

Deep neural networks (DNNs) are vulnerable to adversarial examples, maliciously modified raw input data which is imperceptible to human vision, but once fed into DNNs, can lead DNNs to produce incorrect outputs. The existence and easy construction of adversarial examples pose significant security risks to DNNs, especially in safety-critical applications, including visual object recognition and autonomous driving. The objective of this project is to model adversarial perturbations in DNNs through statistical analysis and DNN visualization. The established model will provide a basis for developing a radically different approach for detecting adversarial examples. 

Supervisor: Prof. En-Hui Yang 
Email: ehyang@uwaterloo.ca 

This project aims to perform numerical simulations of nanoelectronic devices based on nanomaterials, which will explain experimental device performance. 

Required knowledge & skills: Semiconductor device physics

Supervisor: Prof. Na Young Kim
Email: nayoung.kim@uwaterloo.ca
Location: Remote/RAC-2101

Deep neural networks (DNNs) are vulnerable to adversarial examples, maliciously modified raw input data which is imperceptible to human vision, but once fed into DNNs, can lead DNNs to produce incorrect outputs. The existence and easy construction of adversarial examples pose significant security risks to DNNs, especially in safety-critical applications, including visual object recognition and autonomous driving. One way to partially mitigate this problem is to formulate deep learning as a type of minimax problem instead of the standard minimization problem. The objective of this project is to explore and implement effective methods for solving such a minimax problem, yielding robust deep learning.

Supervisor: Prof. En-Hui Yang
Email: ehyang@uwaterloo.ca

While deep learning-based artificial intelligence (AI) has propelled the information age to unprecedented heights, its current trajectory—characterized by scaling up massive deep neural networks (DNNs) and leveraging vast datasets—faces significant limitations. These include enormous demands for energy, computing resources, and high-throughput networking, alongside persistent challenges in AI interpretability and security. Moreover, the high costs associated with this approach restrict accessibility to a few large players, limiting its societal impact and potential to create broader economic opportunities. 

To address these challenges, this project aims to explore and develop alternative AI paradigms that prioritize efficiency, interpretability, security, and inclusivity. By leveraging concepts and techniques from information science and probability theory, the project will initially focus on enhancing the nonlinearity, structural properties, and training/inference efficiencies of DNNs, a critical step towards achieving the final goal. 

Supervisor: Prof. En-Hui Yang 
Email: ehyang@uwaterloo.ca 

Summary:

“A word’s meaning is characterized by the contexts in which it appears.”— Zellig Harris (1954).

Modern large language models (LLMs) such as GPT, LLaMA, and BERT have revolutionized language understanding by embedding Harris's insight into their core: meaning emerges from word co-occurrence and distributional similarity. However, despite their success, current LLMs suffer from a crucial shortcoming—they lack a principled mechanism to compute the actual contextual meaning of a word given its surrounding text.

In these models, the final-layer representation of a word token is optimized for predicting the next token rather than for explicitly capturing the conditional semantics of a word in context. As a result, polysemy resolution, precise semantic interpretation, and context-sensitive word understanding remain unsolved in a strict sense.

To overcome this limitation, we have recently developed a method that directly tackles the problem of extracting the true contextual meaning of a word from its surrounding text. Building on this foundation, this project aims to construct a new generation of LLM-based dictionaries that:

• derives word senses from a semantic universe encoded in high-quality, large-scale written corpora (e.g., FineWeb-Edu);
• clusters, for each word, its usage-based probability distributions to discover the number and content of senses automatically;
• disambiguates the meaning of a word in any given sentence through algorithmic inference, rather than relying on user-selected, human-curated glosses;
• deliver contextually accurate definitions of a word at the point of use—for both human readers and downstream applications

This work not only pushes forward the frontier of dictionary construction but also lays the groundwork for semantic organization and understanding of unstructured textual data.

Supervisor: Prof. En-Hui Yang
Email: ehyang@uwaterloo.ca