Jacob Si

PhD Student at Imperial College London

Hi, I’m Jacob, a Computer Science PhD student at Imperial College London advised by Yingzhen Li and Marek Rei. My research interests span various domains including tabular learning, generative models, variational inference, language agents and post-training alignment. Previously, I was a Research Intern at Samsung AI Research (2025).

Before joining Imperial, I grew up in Singapore and Malaysia, then earned my Bachelor’s from the University of Toronto and my Master’s from the University of California, Los Angeles. I was fortunate to be advised by Rahul Krishnan at the Vector Institute, and Jonathan Kao at UCLA SEAS. Lastly, I am grateful to my instructors, Jimmy Ba and David Duvenaud, for sparking my initial interest in machine learning.

If you are interested in research or would like to collaborate, please feel free to email me!

Email:

‘y.si23‘ @ ‘imperial.ac.uk‘

‘jacobyhsi‘ @ ‘ucla.edu‘

‘jacobyhsi‘ @ ‘cs.toronto.edu‘

Selected Publications [full list]

(*) denotes equal contribution

NeurIPS

Variational Uncertainty Decomposition for In-Context Learning

I. Shavindra Jayasekera*, Jacob Si*, Filippo Valdettaro, Wenlong Chen, Aldo Faisal, and Yingzhen Li

In the 39th Conference on Neural Information Processing System, 2025.

Abstract PDF Code

As large language models (LLMs) gain popularity in conducting prediction tasks in-context, understanding the sources of uncertainty in in-context learning becomes essential to ensuring reliability. The recent hypothesis of in-context learning performing predictive Bayesian inference opens the avenue for Bayesian uncertainty estimation, particularly for decomposing uncertainty into epistemic uncertainty due to lack of in-context data and aleatoric uncertainty inherent in the in-context prediction task. However, the decomposition idea remains under-explored due to the intractability of the latent parameter posterior from the underlying Bayesian model. In this work, we introduce a variational uncertainty decomposition framework for in-context learning without explicitly sampling from the latent parameter posterior, by optimising auxiliary inputs as probes to obtain an upper bound to the aleatoric uncertainty of an LLM’s in-context learning procedure. Through experiments on synthetic and real-world tasks, we show quantitatively and qualitatively that the decomposed uncertainties obtained from our method exhibit desirable properties of epistemic and aleatoric uncertainty.
ICMLSpotlight

InterpreTabNet: Distilling Predictive Signals from Tabular Data by Salient Feature Interpretation

Jacob Si, Wendy Yusi Cheng*, Michael Cooper*, and Rahul Krishnan

In the 41st International Conference on Machine Learning, 2024.
Spotlight Presentation [top 3.5%]
Abstract PDF Code

Tabular data are omnipresent in various sectors of industries. Neural networks for tabular data such as TabNet have been proposed to make predictions while leveraging the attention mechanism for interpretability. However, the inferred attention masks are often dense, making it challenging to come up with rationales about the predictive signal. To remedy this, we propose InterpreTabNet, a variant of the TabNet model that models the attention mechanism as a latent variable sampled from a Gumbel-Softmax distribution. This enables us to regularize the model to learn distinct concepts in the attention masks via a KL Divergence regularizer. It prevents overlapping feature selection by promoting sparsity which maximizes the model’s efficacy and improves interpretability to determine the important features when predicting the outcome. To assist in the interpretation of feature interdependencies from our model, we employ a large language model (GPT-4) and use prompt engineering to map from the learned feature mask onto natural language text describing the learned signal. Through comprehensive experiments on real-world datasets, we demonstrate that InterpreTabNet outperforms previous methods for interpreting tabular data while attaining competitive accuracy.
TMLR

TabRep: Training Tabular Diffusion Models with a Simple and Effective Continuous Representation

Jacob Si, Zijing Ou*, Mike Qu*, Zhengrui Xiang*, and Yingzhen Li

In the Transactions on Machine Learning Research, 2026.

Abstract PDF Code

Diffusion models have been the predominant generative model for tabular data generation. However, they face the conundrum of modeling under a separate versus a unified data representation. The former encounters the challenge of jointly modeling all multi-modal distributions of tabular data in one model. While the latter alleviates this by learning a single representation for all features, it currently leverages sparse suboptimal encoding heuristics and necessitates additional computation costs. In this work, we address the latter by presenting TabRep, a tabular diffusion architecture trained with a unified continuous representation. To motivate the design of our representation, we provide geometric insights into how the data manifold affects diffusion models. The key attributes of our representation are composed of its density, flexibility to provide ample separability for nominal features, and ability to preserve intrinsic relationships. Ultimately, TabRep provides a simple yet effective approach for training tabular diffusion models under a continuous data manifold. Our results showcase that TabRep achieves superior performance across a broad suite of evaluations. It is the first to synthesize tabular data that exceeds the downstream quality of the original datasets while preserving privacy and remaining computationally efficient.
Workshop

TabRAG: Improving Tabular Document Question Answering for Retrieval Augmented Generation via Structured Representations

Jacob Si*, Mike Qu*, Michelle Lee, Marek Rei, and Yingzhen Li

In the 39th Conference on Neural Information Processing System, AI for Tabular Data Workshop, 2025.

Abstract PDF Code

Incorporating external knowledge bases in traditional retrieval-augmented generation (RAG) relies on parsing the document, followed by querying a language model with the parsed information via in-context learning. While effective for text-based documents, question answering on tabular documents often fails to generate plausible responses. Standard parsing techniques lose the two-dimensional structural semantics critical for cell interpretation. In this work, we present TabRAG, a parsing-based RAG framework designed to improve tabular document question answering via structured representations. Our framework consists of layout segmentation that decomposes the document inputs into a series of components, enabling fine-grained extraction. Subsequently, a vision language model parses and extracts the document tables into a hierarchically structured representation. In order to cater various table styles and formats, we integrate a self-generated in-context learning module that guides the table extraction process. Experimental results demonstrate that TabRAG outperforms existing popular parsing techniques across a broad suite of evaluation and ablation benchmarks.
Submission

Diffusion Alignment Beyond KL: Variance Minimisation as Effective Policy Optimiser

Zijing Ou, Jacob Si, Junyi Zhu, Ondrej Bohdal, Mete Ozay, Taha Ceritli, and Yingzhen Li

In submission.

Abstract PDF

Diffusion alignment adapts pretrained diffusion models to sample from reward-tilted distributions along the denoising trajectory. This process naturally admits a Sequential Monte Carlo (SMC) interpretation, where the denoising model acts as a proposal and reward guidance induces importance weights. Motivated by this view, we introduce Variance Minimisation Policy Optimisation (VMPO), which formulates diffusion alignment as minimising the variance of log importance weights rather than directly optimising a Kullback-Leibler (KL) based objective. We prove that the variance objective is minimised by the reward-tilted target distribution and that, under on-policy sampling, its gradient coincides with that of standard KL-based alignment. This perspective offers a common lens for understanding diffusion alignment. Under different choices of potential functions and variance minimisation strategies, VMPO recovers various existing methods, while also suggesting new design directions beyond KL.
Book Chapter

Assessing Infant Mortality Rate: Problems stemming from Household Living Conditions, Women’s Education and Health

Jacob Si, and Rohan Alexander

In "Telling Stories with Data: With Applications in R" by Rohan Alexander

Abstract PDF

What areas can be improved in order to promote the well-being of women in India and hence, reduce the infant mortality rate? Utilizing the data from the 1998-1999 India National Family Health Survey provided by the Demographic and Health Survey (DHS) program, we look to depict the demographics of Indian women and infants in different states of India. We have found that the root causes of poor infant mortality rates stem from having poor living conditions that affect the likelihood of women to attain education and understand the importance of antenatal care and birth delivery assistance. We also explore other factors such as potentially inheritable traits (unhealthy body weight and anaemia disease) as well as an infant’s diet. These factors are crucial in the development of an infant and the reduction of the infant mortality rate.