A new DDPM for link prediction using sub-graph likelihood estimation.
problem Link prediction in graph domains.
method Sub-graph based diffusion model with DDPMs, decomposing likelihood estimation.
result Our model achieves superior performance in link prediction across various datasets.
LGCL uses learnable filters to apply CNNs on graphs.
problem Applying CNNs to generic graphs like networks.
method Learnable graph convolutional layer (LGCL) and sub-graph training.
result LGCL achieves better performance on various graph datasets.
This paper extracts topic-specific subgraphs from Wikidata for easier research.
problem Time-consuming and task-specific tuning in using Wikidata for graph processing.
method Unified framework to extract topic-specific subgraphs from Wikidata.
result Standardized sub-graphs for easier evaluation of graph processing algorithms.
Characterizes geometric actions on graphs with flexible stabilizers.
problem Understanding geometric actions on flexible stabilizers.
method Defining generalized fine actions and proving relative quasi-convexity criteria.
result Characterizes Bowditch boundary points in relatively geometric actions.
NAS-Navigator automates neural network architecture search with visual steering.
problem Difficulty in configuring large neural networks efficiently.
method Formulates architecture optimization as graph space exploration, trains all candidate architectures in one-shot.
result Allows analysts to effectively select and guide the search for optimal neural network architectures.
Given a hyperbolic surface, the set of all closed geodesics whose length is minimal form a graph on the surface, in fact a so-called fat graph, which we call the systolic graph. We study which fat graphs are systolic graphs for some surface (we call these admissible). There is a natural necessary condition on such grap…
We explore several families of flip-graphs, all related to polygons or punctured polygons. In particular, we consider the topological flip-graphs of once-punctured polygons which, in turn, contain all possible geometric flip-graphs of polygons with a marked point as embedded sub-graphs. Our main focus is on the geometr…
New method reduces errors in causal discovery from data.
problem Errors in causal discovery from limited data.
method Hierarchical wrapper for constraint-based algorithms.
result Significantly fewer tests, more accurate graphs, shorter run-times.
A key challenge for Bitcoin cryptocurrency holders, such as startups using ICOs to raise funding, is managing their FX risk. Specifically, a misinformed decision to convert Bitcoin to fiat currency could, by itself, cost USD millions. In contrast to financial exchanges, Blockchain based crypto-currencies expose the ent…
DAGCN improves graph classification by learning neighbor importance and pooling.
problem Loss of early-stage information and loss of node characteristics in GCNs.
method Dual attention graph convolution and self-attention pooling.
result DAGCN outperforms state-of-the-art methods in graph classification.
Proves accuracy guarantees for self-supervised learning with correlated positive pairs.
problem Lack of theoretical guarantees for self-supervised learning with correlated positive pairs.
method Novel augmentation graph concept and spectral decomposition loss.
result Provably accurate features under linear probe evaluation.
ECG improves graph clustering and resolves resolution limit issues.
problem Graph clustering resolution limit issue.
method ECG uses consensus clustering to improve graph clustering.
result ECG alleviates the resolution limit issue and improves partition stability.
Deep neural networks (DNNs) have become the state-of-the-art technique for machine learning tasks in various applications. However, due to their size and the computational complexity, large DNNs are not readily deployable on edge devices in real-time. To manage complexity and accelerate computation, network compression…
In this thesis, we propose several modelling strategies to tackle evolving data in different contexts. In the framework of static clustering, we start by introducing a soft kernel spectral clustering (SKSC) algorithm, which can better deal with overlapping clusters with respect to kernel spectral clustering (KSC) and p…
Graph neural networks are shown to be as powerful as a graph isomorphism heuristic, leading to a new approach for higher-order graph structures.
problem Understanding and distinguishing non-isomorphic graphs and their higher-order structures.
method Relating GNNs to the Weisfeiler-Leman heuristic and proposing k-dimensional GNNs. result GNNs have the same expressiveness as the Weisfeiler-Leman heuristic in distinguishing graphs and their higher-order structures.
Proposes an evolutionary approach to fitting acyclic VAR models.
problem Cycles in multivariate time series systems obscure hierarchical analysis.
method Evolutionary approach to fitting acyclic VAR processes with hierarchical representation.
result Outperforms unconstrained models and captures key structural properties.
Boosts share routing for multi-task learning with flexible sparse connections.
problem Designing suitable sharing mechanisms among multiple tasks in multi-task learning.
method Proposes MTNAS framework to modularize sharing into sub-networks with sparse connections and gating.
result Demonstrates consistent improvement over single-task and typical multi-task methods while maintaining efficiency.
COMBO optimizes Bayesian Optimization for combinatorial search spaces.
problem Optimizing objectives on combinatorial search spaces with high-order interactions.
method COMBO uses a combinatorial graph and ARD diffusion kernel with Horseshoe prior for efficient modeling and variable selection.
result COMBO outperforms state-of-the-art methods consistently across various benchmarks.
Transformer adapts to graphs with adaptive attention and auto-regressive decoding.
problem Transformers struggle with graph data due to non-sequential nature.
method Proposes GRAT, a Transformer variant with adaptive attention and auto-regressive decoding.
result GRAT achieves state-of-the-art performance on molecule property predictions and generation tasks.
Graph Neural Network identifies ASD biomarkers from fMRI data.
problem Finding biomarkers for Autism Spectrum Disorder (ASD).
method Graph Neural Network (GNN) for analyzing task-fMRI brain networks, 2-stage pipeline to interpret feature importance.
result GNN achieves high accuracy in identifying ASD biomarkers and reveals their association with social behaviors.
A new method for identifying causal directions in complex systems.
problem Identifying causal relationships in nonlinear systems with limited data.
method Sequential edge orientation approach using pairwise additive noise model.
result The method can recover true causal DAGs under nonlinear additive noise models.
The paper models reciprocity in interbank markets using a statistical null model.
problem Understanding the importance of individual banks in financial networks.
method Developed an exponential random graph model to account for reciprocal links on both topological and weighted levels.
result Weighted reciprocity in interbank markets is more significant than network size and volume before the financial crisis.
This study diversifies stock and crypto portfolios using network analysis.
problem Balancing returns and volatility in diversified portfolios.
method Community detection in network representations of assets, using Louvain and Affinity propagation algorithms.
result Opposite trends in crypto and traditional asset markets.
Paper recovers lattice signal partitions efficiently.
problem Estimating lattice partition from noisy data.
method Uses dyadic CART for computationally-efficient partition recovery.
result Consistently estimates partition with optimal error rate.
We fill a void in merging empirical and phenomenological characterisation of the dynamical phase transitions in complex systems by identifying three of them on real-life financial markets. We extract and interpret the empirical, numerical, and semi-analytical evidences for the existence of these phase transitions, by c…
Previously in 2014, we proposed the Nearest Descent (ND) method, capable of generating an efficient Graph, called the in-tree (IT). Due to some beautiful and effective features, this IT structure proves well suited for data clustering. Although there exist some redundant edges in IT, they usually have salient features …
Subg-Con learns graph representations from subgraphs, improving scalability and efficiency.
problem Scalability issues and weak supervision in graph representation learning.
method Subg-Con uses subgraphs sampled from the original graph to define a contrastive loss, learning node representations without complete graph data.
result Subg-Con outperforms existing methods in scalability, efficiency, and weak supervision requirements.
We prove a discrete Jordan-Brouwer-Schoenflies separation theorem telling that a (d-1)-sphere H embedded in a d-sphere G defines two different connected graphs A,B in G such a way that the intersection of A and B is H and the union is G and such that the complementary graphs A,B are both d-balls. The graph theoretic de…
Taylorized training improves neural network training at finite width.
problem Understanding and improving neural network training at finite width.
method Training the k-th order Taylor expansion of the neural network at initialization.
result Taylorized training agrees with full neural network training better as k increases and can significantly close the performance gap.
Self-training outperforms pre-training on COCO object detection and segmentation datasets.
problem The effectiveness of pre-training in improving object detection and segmentation models is limited.
method Investigated self-training as an alternative method to utilize additional data.
result Self-training consistently improves model performance across various dataset sizes and data augmentation levels.
Un-trained neural networks outperform trained methods in MRI reconstruction.
problem Accelerated MRI reconstruction with minimal training data.
method Variation of Deep Decoder without training data.
result Un-trained approach significantly outperforms other methods in reconstruction accuracy.
Meta-learning performance depends on train-validation split type.
problem Understanding the importance of train-validation split in meta-learning.
method Theoretical and experimental study comparing train-val and train-train methods.
result Train-train method can achieve strictly better excess loss in realizable cases.
New method reveals how training data influence diffusion model outputs.
problem Difficulty in assessing training data impact on diffusion model outputs.
method Use of ensembles trained on carefully engineered splits of training data to identify influential training examples.
result Demonstrated the viability of ensembles as generative models and validity of assessing influence.
Paper shows adversarial training can be fooled by new type of noise.
problem Adversarial training can be fooled by new types of noise.
method Designing ADVIN, a new type of inducing noise.
result ADVIN can degrade adversarial training robustness by 99.9%.
New MIP methods improve training of integer-valued neural networks.
problem Training integer-valued neural networks with limited data and resources.
method Formulated new MIP models to optimize training efficiency and handle more data.
result Significantly outperforms previous state-of-the-art methods in accuracy, training time, and data usage.
Free adversarial training reduces the generalization gap compared to vanilla method.
problem Improving generalization in adversarial training.
method Analysis of algorithmic stability in free adversarial training.
result Free adversarial training shows a lower generalization gap.
ADASS selects adaptive subsets for SGD training acceleration.
problem Fixed sample size in SGD limits training efficiency.
method ADASS selects adaptive subsets based on Lipschitz constants.
result ADASS achieves comparable accuracy with full training set.
Solve-training trains neural nets to map physical solutions efficiently.
problem Representing complex physical solutions with neural networks.
method Variational training using loss functions from physical models.
result Effective neural network representation of solution maps without expensive labels.
MixTrain improves verifiable robustness of neural networks without sacrificing efficiency.
problem Efficiently making neural networks robust against adversarial attacks.
method Stochastic robust approximation and dynamic mixed training techniques.
result Achieves up to 95.2% verified robust accuracy with significantly reduced training time.
learn2mix trains neural nets faster by adjusting class proportions dynamically.
problem Training neural nets efficiently with limited resources and imbalanced classes.
method Adaptive class proportion adjustment during training.
result Neural nets trained with learn2mix converge faster than static methods.
Loss-guided training accelerates node embedding methods on graphs.
problem Training efficiency in graph learning methods with implicit positive examples.
method Dynamic adjustment of training distribution based on loss values.
result Significant acceleration in training and computation over static methods.
Paper explores fast adversarial training to improve robustness with less computation.
problem Efficiently defending against adversarial examples.
method Integrates simple self-attacks for faster training, focusing on overfitting recovery.
result Shows superior robust accuracy with reduced training time compared to strong adversarial training.
Three LF training criteria improve neural network acoustic models without cross-entropy pre-training.
problem Improving purely sequence-trained neural network acoustic models.
method Comparison of three lattice-free discriminative training criteria (MMI, bMMI, sMBR) on LVCSR tasks.
result LF-bMMI models outperform plain LF-MMI models by 5% WER on Switchboard datasets.
Free adversarial training improves robustness without generating adversarial examples.
problem Training robust models against adversarial attacks is costly and impractical for large-scale datasets.
method Recycles gradient information from parameter updates to generate adversarial examples.
result Free adversarial training achieves comparable robustness to PGD training at negligible cost.
Detects backdoors in trained models without poisoned training data.
problem Detecting backdoors in DNNs trained without access to the poisoned training set.
method Proposes a novel detector using the maximum achievable misclassification fraction (MAMF) statistic.
result Detects backdoors and infers source and target classes.
Pre-training improves model robustness and uncertainty.
problem Improving model robustness and uncertainty in machine learning.
method Adversarial pre-training and task-specific methods.
result Approximately a 10% absolute improvement in adversarial robustness.
Crowdsourced training of large neural networks with decentralized Mixture-of-Experts.
problem Expensive training of large neural networks limits research contributions.
method Learning@home: decentralized Mixture-of-Experts for large, poorly connected participants.
result Performance and reliability of Learning@home surpass conventional distributed training.
Pre-training is crucial for learning deep neural networks. Most of existing pre-training methods train simple models (e.g., restricted Boltzmann machines) and then stack them layer by layer to form the deep structure. This layer-wise pre-training has found strong theoretical foundation and broad empirical support. Howe…