25 articles tagged with #few-shot-learning. AI-curated summaries with sentiment analysis and key takeaways from 50+ sources.
Researchers introduce BiCLIP, a new framework that improves vision-language models' ability to adapt to specialized domains through geometric transformations. The approach achieves state-of-the-art results across 11 benchmarks while maintaining simplicity and low computational requirements.
Researchers introduce TATRA, a training-free prompting method for Large Language Models that creates instance-specific few-shot prompts without requiring labeled training data. The method achieves state-of-the-art performance on mathematical reasoning benchmarks like GSM8K and DeepMath, matching or outperforming existing prompt optimization methods that rely on expensive training processes.
Researchers introduce RDB-PFN, the first relational foundation model for databases trained entirely on synthetic data to overcome privacy and scarcity issues with real relational databases. The model uses a Relational Prior Generator to create over 2 million synthetic tasks and demonstrates strong few-shot performance on 19 real-world relational prediction tasks through in-context learning.
Researchers introduce SPRINT, the first Few-Shot Class-Incremental Learning (FSCIL) framework designed specifically for tabular data domains like cybersecurity and healthcare. The system achieves 77.37% accuracy in 5-shot learning scenarios, outperforming existing methods by 4.45% through novel semi-supervised techniques that leverage unlabeled data and confidence-based pseudo-labeling.
Researchers propose Supervised Calibration (SC), a new framework to improve In-Context Learning performance in Large Language Models by addressing systematic biases through optimal affine transformations in logit space. The method achieves state-of-the-art results across multiple LLMs including Mistral-7B, Llama-2-7B, and Qwen2-7B in few-shot learning scenarios.
Researchers developed an energy-based AI model that can learn spatial concepts like 'near' and 'above' from just five demonstrations using 2D point sets. The model demonstrates cross-domain transfer capabilities, applying concepts learned in 2D particle environments to solve 3D physics-based robotics tasks.
Researchers introduce Legal2LogicICL, an LLM-based framework that improves the conversion of natural-language legal cases into logical formulas through retrieval-augmented few-shot learning. The method addresses data scarcity in legal AI systems and introduces a new annotated dataset (Legal2Proleg) to advance interpretable legal reasoning without requiring model fine-tuning.
Researchers present the first comprehensive survey of inductive reasoning in large language models, categorizing improvement methods into post-training, test-time scaling, and data augmentation approaches. The survey establishes unified benchmarks and evaluation metrics for assessing how LLMs perform particular-to-general reasoning tasks that better align with human cognition.
Researchers propose GNN-as-Judge, a framework combining Large Language Models with Graph Neural Networks to improve learning on text-attributed graphs in low-resource settings. The approach uses collaborative pseudo-labeling and weakly-supervised fine-tuning to generate reliable labels while reducing noise, demonstrating significant performance gains when labeled data is scarce.
Researchers introduce Uni-DAD, a unified approach that combines diffusion model distillation and adaptation into a single pipeline for efficient few-shot image generation. The method achieves comparable quality to state-of-the-art methods while requiring less than 4 sampling steps, addressing the computational cost issues of traditional diffusion models.
Researchers propose Q-LoRA, a quantum-enhanced fine-tuning method that integrates quantum neural networks into LoRA adapters for improved AI-generated content detection. The study also introduces H-LoRA, a classical variant using Hilbert transforms that achieves similar 5%+ accuracy improvements over standard LoRA at lower computational cost.
Researchers developed a meta-learning approach for Large Multimodal Models (LMMs) that uses distilled soft prompts to improve few-shot visual question answering performance. The method outperformed traditional in-context learning by 21.2% and parameter-efficient finetuning by 7.7% on VQA tasks.
The article discusses advancements in time series foundation models and their capability for few-shot learning in generative AI applications. These models can learn patterns from limited data samples, potentially improving forecasting and prediction tasks across various domains.
SetFit is a new machine learning framework that enables efficient few-shot learning without requiring prompts. This approach could significantly reduce the computational resources and data requirements for training AI models in various applications.
Meta reinforcement learning enables AI agents to rapidly adapt to new tasks by learning from a distribution of training tasks. The approach allows agents to develop new RL algorithms through internal activity dynamics, focusing on fast and efficient problem-solving for unseen scenarios.
Researchers propose directional CDNV (decision-axis variance) as a key geometric quantity explaining why self-supervised learning representations transfer well with few labels. The study shows that small variability along class-separating directions enables strong few-shot transfer and low interference across multiple tasks.
Researchers propose a Label-guided Distance Scaling (LDS) strategy to improve few-shot text classification by leveraging label semantics during both training and testing phases. The method addresses misclassification issues when randomly selected labeled samples don't provide effective supervision signals, demonstrating significant performance improvements over state-of-the-art models.
Researchers have created CrimeNER, a specialized dataset of over 1,500 annotated crime-related documents for training named-entity recognition AI models. The study addresses the lack of quality training data in the crime domain by developing a database from terrorist attack reports and DOJ press notes, defining 22 types of crime-related entities.
Researchers propose a Manifold Residual (MR) block to address overfitting in few-shot Whole Slide Image classification by preserving the low-dimensional manifold geometry of pathology foundation model features. The geometry-aware approach achieves state-of-the-art results with fewer parameters by using a fixed random matrix as geometric anchor and a trainable low-rank residual pathway.
MAGIC is a new AI framework for few-shot anomaly detection in industrial quality control that uses mask-guided inpainting to generate high-fidelity synthetic anomalies. The system introduces three key innovations: Gaussian prompt perturbation, spatially adaptive guidance, and context-aware mask alignment to improve anomaly generation while preserving normal regions.
Google's Gemini AI is being trained to identify exploding stars (supernovas) using few-shot learning techniques. This demonstrates AI's capability to recognize rare astronomical phenomena with minimal training examples.
The article appears to discuss GPT-Neo and Hugging Face's Accelerated Inference API in the context of few-shot learning applications. However, the article body content is empty, preventing detailed analysis of the technical implementation or market implications.
Researchers introduce MAML-KT, a meta-learning approach that addresses the cold start problem in knowledge tracing systems when predicting performance of new students with limited interaction data. The model uses few-shot learning to rapidly adapt to unseen students, achieving higher early accuracy than existing knowledge tracing models across multiple datasets.
The article appears to discuss SetFitABSA, a methodology for performing aspect-based sentiment analysis using SetFit with minimal training examples. However, the article body is empty, making it impossible to provide meaningful analysis of the content or implications.
The article title references few-shot learning capabilities in language models, but no article body content was provided for analysis. Without the actual article content, a comprehensive analysis cannot be performed.
