A5094008624- Neural Networks and Applications
- Multimodal Machine Learning Applications
- Gaussian Processes and Bayesian Inference
- Topic Modeling
- Natural Language Processing Techniques
- Advanced Neural Network Applications
- Scientific Computing and Data Management
- Time Series Analysis and Forecasting
- Domain Adaptation and Few-Shot Learning
- Biomedical Text Mining and Ontologies
KTH Royal Institute of Technology
2024
Time Series Representation Learning (TSRL) focuses on generating informative representations for various (TS) modeling tasks. Traditional Self-Supervised (SSL) methods in TSRL fall into four main categories: reconstructive, adversarial, contrastive, and predictive, each with a common challenge of sensitivity to noise intricate data nuances. Recently, diffusion-based have shown advanced generative capabilities. However, they primarily target specific application scenarios like imputation...
Multimodal Large Language Models (MLLMs) are commonly evaluated using costly annotated multimodal benchmarks. However, these benchmarks often struggle to keep pace with the rapidly advancing requirements of MLLM evaluation. We propose GenCeption, a novel and annotation-free evaluation framework that merely requires unimodal data assess inter-modality semantic coherence inversely reflects models' inclination hallucinate. Analogous popular DrawCeption game, GenCeption initiates non-textual...
Time Series Representation Learning (TSRL) focuses on generating informative representations for various (TS) modeling tasks. Traditional Self-Supervised (SSL) methods in TSRL fall into four main categories: reconstructive, adversarial, contrastive, and predictive, each with a common challenge of sensitivity to noise intricate data nuances. Recently, diffusion-based have shown advanced generative capabilities. However, they primarily target specific application scenarios like imputation...
Tabular synthesis models remain ineffective at capturing complex dependencies, and the quality of synthetic data is still insufficient for comprehensive downstream tasks, such as prediction under distribution shifts, automated decision-making, cross-table understanding. A major challenge lack prior knowledge about underlying structures high-order relationships in tabular data. We argue that a systematic evaluation on structural information first step towards solving problem. In this paper,...