Research

My primary interest lies in deep learning, but I also maintain a broad set of interests within machine learning. In particular, I am interested in analyzing neural network models under the view of dynamical system or probabilistic perspective.



Papers

2023
Onchain Sports Betting using UBET Automated Market Maker

Daniel Jiwoong Im, Alexander Kondratskiy, Vincent Harvey, Hsuan-Wei Fu
Technical script - whitepaper for UBET Sports 2023
[PDF]


UAMM: UBET Automated Market Maker

Daniel Jiwoong Im, Alexander Kondratskiy, Vincent Harvey, Hsuan-Wei Fu
https://arxiv.org/abs/2308.06375 2023
[PDF]


Active and Passive Causal Inference Learning

Daniel Jiwoong Im, Kyunghyun Cho
https://arxiv.org/abs/2308.09248 2023
[PDF]


2021
Causal Effect Variational Autoencoder with Uniform Treatment

Daniel Jiwoong Im, Kyunghyun Cho, Narges Razavian
https://arxiv.org/abs/2111.08656 2021
[PDF]


Online hyperparameter optimization by real-time recurrent learning

Daniel Jiwoong Im, Cristina Savin, Kyunghyun Cho
https://arxiv.org/abs/2102.07813 2021
[PDF, Code]


2020
Evaluation metrics for behaviour modeling

Daniel Jiwoong Im, Iljung Kwak, Kristin Branson
https://arxiv.org/abs/2007.12298 2020
[PDF, Presentation]


2019
Are skip connections necessary for biologically plausible learning rules?

Daniel Jiwoong Im, Rutujia Patil, Kristin Branson
Neural Information Processing Systems (NeuralIPs) Neuro AI Workshop 2019
[PDF, Poster]


Model-Agnostic Meta-Learning using Runge-Kutta Methods

Daniel Jiwoong Im, Yibo Jiang, Nakul Verma
https://arxiv.org/abs/1910.07368 2019
[PDF]


Importance Weighted Adversarial Variational Autoencoders for Spike Inference from Calcium Imaging Data

Daniel Jiwoong Im, Sridhama Prakhya, Jinyao Yan, Srinivas Turaga, Kristin Branson
https://arxiv.org/abs/1906.03214 2019
[PDF]


2018
Stochastic Neighbor Embedding under f-divergences

Daniel Jiwoong Im, Nakul Verma, Kristin Branson
https://arxiv.org/abs/1811.01247 2018
[PDF]


Quantitatively Evaluating GANs with Divergence Proposed for Training

Daniel Jiwoong Im, Allan He Ma, Graham Taylor, Kristin Branson
International Conference on Learning Representation (ICLR) 2018
[PDF, OpenReview, Poster]


Neural Machine Translation with Gumbel-Greedy Decoding

Jiatao Gu, Daniel Jiwoong Im, Victor O.K. Li
AAAI-18: Thirtieth AAAI Conference on Artificial Intelligence (AAAI) 2018
[PDF]


2017
Denoising Criterion for Variational Auto-encoding Framework

Daniel Jiwoong Im, Sungjin Ahn, Roland Memisevic, Yoshua Bengio
AAAI-17: Thirtieth AAAI Conference on Artificial Intelligence (AAAI) 2017
Arxiv version : http://arxiv.org/abs/1511.06406 2016
[PDF, Poster, Code]


2016
Generative Adversarial Parallelization

Daniel Jiwoong Im, He Ma, Chris Dongjoo Kim, Graham Taylor
https://arxiv.org/abs/1612.04021 2016
[PDF]
An Empirical Analysis of Deep Network Loss Surfaces

Daniel Jiwoong Im, Michael Tao, Kristin Branson
http://arxiv.org/abs/1602.05110 2016
[PDF, Summary]
Learning a Metric for Class-Conditional KNN

Daniel Jiwoong Im, Graham W. Taylor
Internaltional Joint Conference on Neural Networks (IJCNN) 2016 (oral)
[PDF, Slides, Code]


Generating images with recurrent adversarial networks

Daniel Jiwoong Im, Chris Dongjoo Kim, Hui Jiang, Roland Memisevic
http://arxiv.org/abs/1602.05110
[PDF, Code]
Earlier version:Generative Adversarial Metric
Appeared at International Conference on Learning Representations (ICLR) workshop track 2016 [Poster]


Conservativeness of untied auto-encoders

Daniel Jiwoong Im, Mohamed Ishmael Diwan Belghazi, Roland Memisevic
AAAI-16: Thirtieth AAAI Conference on Artificial Intelligence (AAAI) 2016
[PDF, Supplementary Material, Code]


2015
An Empirical Investigation of Minimum Probability Flow Learning under Different Connection Patterns

Daniel Jiwoong Im, Ethan Buchman, and Graham W. Taylor
European Conference of Machine Learning (ECML PKDD) (oral) 2015
[Preprint, Background, Code]
Earlier version: Understanding Minimum Probability Flow for RBMs Under Various Kinds of Dynamics
Appeared at International Conference on Learning Representations (ICLR) workshop track 2015 [PDF, Poster]


Scoring and Classifying with Gated Auto-encoders

Daniel Jiwoong Im, and Graham W. Taylor
European Conference of Machine Learning (ECML PKDD) (oral) 2015
[PDF, Supplementary Material]
Earlier version: Analyzing the Dynamics of Gated Auto-encoders
Appeared at Neural Information Processing Systems Deep Learning Workshop 2014 [PDF, Poster]


Semi-supervised Hyperspectral Image Classification via Neighbourhood Graph Learning

Daniel Jiwoong Im, and Graham W. Taylor
IEEE Geoscience and Remote Sensing Letters 2015 [Link, Preprint]
 Earlier version: Improving semi-supervised neural networks for scene understanding by learning the neighborhood graph
Appeared at Computer Vision and Pattern Recognition Scene Understanding Workshop. 2014 [PDF, Poster]


Analyzing Unsupervised Representation Learning Models Under the View of Dynamical Systems

Master's Thesis, Daniel Jiwoong Im [PDF]


 2014
Neural Network Regularization via Robust Weight Factorization

Jan Rudy, Weiguan Ding, Daniel Jiwoong Im, and Graham W. Taylor
arxiv preprint [PDF]


 2013
How Is Math Applied In Finance

Jiwoong Im, Notes from the Margin Volume VI 2013, Press: Canadian Mathematical Society 2013, 4-5, Print.
[PDF]


 An Empirical Evaluation of the Numerical Techiques on American Put Option Valuation

Jiwoong Im, Project report, 2013
[PDF, Poster]