Training & Fine-tuning A
Showing 91–99 of 99
-
ClinHallu: A Benchmark for Diagnosing Stage-Wise Hallucinations in Medical MLLM ReasoningClinHallu: a stage-wise hallucination diagnosis benchmark for medical MLLMsClinHallu is a benchmark for diagnosing where hallucinations originate in medical multimodal LLM reasoning, decomposing traces into visual recognition, knowledge recall, and reasoning integration. It provides 7,031 validated instances and uses stage-replacement interventions to localize error sources.
-
Graph Structured Combinatorial Semi-Bandit with Nonlinear Reward Associations through Separable SignalsGraph-structured combinatorial semi-bandits with nonlinear rewardsThe paper addresses combinatorial semi-bandit identification of optimal structures under nonlinear reward associations. It leverages separable signals to reduce sampling and computational cost.
-
From Self-Supervised Speech Models to Mixture-of-Experts for Robust Anti-SpoofingSelf-supervised speech models plus MoE for robust anti-spoofingAdvances in speech generation make synthetic speech more natural and spoofing detection harder. The paper combines self-supervised speech models with a mixture-of-experts design to build more robust anti-spoofing systems.
-
When Good Verifiers Go Bad: Self-Improving VLMs Can Regress on New TasksWhen good verifiers go bad: self-improving VLMs can regress on new tasksVerifier-driven self-DPO, where a frozen verifier scores candidates to form preference pairs, is a common recipe for self-improving vision-language models. The paper shows that under this setup VLMs can regress on new tasks when the verifier misbehaves.
-
A Comparative Study of Deep Learning Architectures for Multi-Horizon Behavioural Forecasting for Mobile HealthComparing deep learning for multi-horizon behavioural forecasting in mHealthWearables and smartphones generate rich behavioural time series for proactive health interventions, yet systematic comparisons of forecasting architectures are lacking. The paper benchmarks deep learning architectures for multi-horizon behavioural forecasting in mobile health.
-
Cluster LOCO: Feature Importance For Interpreting ClustersCluster LOCO gives feature importance to interpret clustersClustering is widely used but its outputs are hard to interpret and audit. Cluster LOCO provides feature-importance scores to explain what distinguishes each cluster.
-
BayLing-Duplex: Native Full-Duplex Speech Dialogue with a Single Autoregressive LLMBayLing-Duplex: native full-duplex speech dialogue from one LLMBayLing-Duplex enables native full-duplex speech interaction with a single autoregressive LLM, letting it listen and speak simultaneously. It handles natural phenomena such as overlap and hesitation.
-
Dense Coordinate-List Fine-Tuning Induces a Controllable Interference Surface in Vision-Language ModelsDense coordinate-list fine-tuning induces a controllable interference surfaceFine-tuning vision-language models to emit dense coordinate lists improves grounding but alters how they serialize, repeat, and terminate structured output. The paper shows this induces a controllable interference surface in VLMs.
-
A Fixed-Point Neural Operator for Size- and Functional-Transferable Hamiltonian PredictionA fixed-point neural operator for transferable Hamiltonian predictionPredicting the Kohn-Sham Hamiltonian with ML can accelerate density functional theory while retaining orbitals and energy levels. The paper proposes a fixed-point neural operator for size- and functional-transferable Hamiltonian prediction.