COMP6258: Differentiable Programming and Deep Learning

2024-25


Maintained by Professor Jonathon Hare and Dr Antonia Marcu.

Deep Learning Coursework: The COMP6258 Reproducibility Challenge - Paper Ideas

URL Title
https://openreview.net/forum?id=6O3Q6AFUTu NoiseDiffusion: Correcting Noise for Image Interpolation with Diffusion Models beyond Spherical Linear Interpolation
https://proceedings.mlr.press/v235/qiao24a.html Ensemble Pruning for Out-of-distribution Generalization
https://proceedings.mlr.press/v235/liu24ar.html Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency
https://proceedings.mlr.press/v235/kim24c.html LayerMerge: Neural Network Depth Compression through Layer Pruning and Merging
https://proceedings.mlr.press/v235/horoi24a.html Harmony in Diversity: Merging Neural Networks with Canonical Correlation Analysis
https://proceedings.mlr.press/v235/gong24b.html Does Label Smoothing Help Deep Partial Label Learning?
https://proceedings.mlr.press/v235/balestriero24b.html How Learning by Reconstruction Produces Uninformative Features for Perception
https://openreview.net/pdf?id=xutrKezbPF CIFD: Controlled Information Flow to Enhance Knowledge Distillation
https://openreview.net/pdf?id=uwSaDHLlYc Diversity-Driven Synthesis: Enhancing Dataset Distillation through Directed Weight Adjustment
https://openreview.net/pdf?id=owuEcT6BTl Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space
https://openreview.net/pdf?id=opt72TYzwZ Optimal ablation for interpretability
https://openreview.net/pdf?id=oNMnR0NJ2e A Label is Worth a Thousand Images in Dataset Distillation
https://openreview.net/pdf?id=lrSrJZZCle CODA: A Correlation-Oriented Disentanglement and Augmentation Modeling Scheme for Better Resisting Subpopulation Shifts
https://openreview.net/pdf?id=IoRT7EhFap Addressing Spectral Bias of Deep Neural Networks by Multi-Grade Deep Learning
https://openreview.net/pdf?id=ebBnKVxMcZ Confidence Calibration of Classifiers with Many Classes
https://openreview.net/pdf?id=96gXvFYWSE Pearls from Pebbles: Improved Confidence Functions for Auto-labeling
https://openreview.net/pdf?id=8HeUvbImKT WeiPer: OOD Detection using Weight Perturbations of Class Projections
https://openreview.net/pdf?id=2TktDpGqNM Overcoming Common Flaws in the Evaluation of Selective Classification Systems
https://openreview.net/forum?id=XmyxQaTyck&noteId=1FUnA4F1N3 Benchmarking the Attribution Quality of Vision Models
https://openreview.net/pdf?id=cmcD05NPKa LEARNING THE GREATEST COMMON DIVISOR: EXPLAINING TRANSFORMER PREDICTIONS
https://openreview.net/forum?id=zlgfRk2CQa Rethinking Deep Thinking: Stable Learning of Algorithms using Lipschitz Constraints
https://openreview.net/forum?id=YrAxxscKM2 Why Do We Need Weight Decay in Modern Deep Learning?
https://openreview.net/forum?id=nxL7eazKBI Model LEGO: Creating Models Like Disassembling and Assembling Building Blocks
https://openreview.net/forum?id=4bKEFyUHT4 Convolutional Differentiable Logic Gate Networks
Copyright ©2025 The University of Southampton. All rights reserved.