The proposed method qualitatively and quantitatively analyzes causal factors using controlled variables and counterfactual interventions
Explore technology by industry and role
Newest additions first. Explore past and current notes by industry and role.
All notes
Newest additions first. Explore past and current notes by industry and role.
Observes complementarity between cropping and compression, and proposes adaptive routing based on input image and question.
A specialized LLM-based system optimized for segment-level quality prediction and error-free segment detection in translation evaluation.
Introduces a multi-view appearance noise prior and super-resolution module to enhance consistency and resolution
A recursive logic with five principles is proposed to improve mechanistic inference in interpretability.
A game-based framework is introduced to explore how LLMs discover research ideas through yes/no questions.
A novel framework integrating 3D point cloud and textual data for open-vocabulary perception
TraIL-Det introduces transformation-invariant local features that capture neighborhood geometry while maintaining robustness to point density variations
A tracking model generates pseudo-ground-truths from LiDAR frame sequences to enhance few-shot learning.
A single representation space exhibits emergent capabilities at functional, residue, and bond levels
A cost-minimized power management model that accounts for battery degradation and operational conditions is proposed.
Custom instructions enable accelerator orchestration and reduced load/store latency for AI workloads.
A universal attack scheme using module-wise noise injection is proposed
The DRIVE framework proposes a solution to improve interpretability consistency and stability in autonomous driving models.
A dual-layer architecture separates routine driving from critical decision-making, mimicking human reasoning.