A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
K.A. Cobbina, Tianyi Zhou
First systematic study of how demo positions in prompts cause accuracy and prediction drift in LLMs, finding start-of-prompt placement yields +6 point gains.
Xiaoquan Chu, Tanlin Sun, Qian Li, et al.
PSPredictor uses machine learning on sequence features to predict liquid-liquid phase separating proteins, achieving 94.71% cross-validation accuracy.
Arwa Bin Raies, Vladimir B. Bajić
This paper provides a comprehensive overview of computational methods for predicting chemical toxicity, comparing their strengths and weaknesses.
Lavender Yao Jiang, Xujin Chris Liu, Nima Pour Nejatian, et al.
NYUTron, a large language model trained on unstructured clinical notes, serves as an all-purpose predictive engine for diverse clinical tasks, outperforming traditional models by 5-15% AUC.
Jonghyun Lee, Dae Won Jun, Ildae Song, et al.
DLM-DTI uses a hint-based learning strategy to create a compact and efficient target encoder for drug-target interaction prediction, reducing VRAM usage to 7.7GB.
Aniruddha Dutta, Saket Kumar, Meheli Basu
This paper demonstrates that a GRU model with recurrent dropout outperforms existing models for Bitcoin price prediction and can yield financial gains when used in simple trading strategies.
Graves, Alex
Proposes an LSTM-based recurrent neural network for denoising and predicting GNSS time series, achieving 50% noise reduction and 1.1 mm MSE prediction error.
Evan Shelhamer, Jonathan Long, Trevor Darrell
Fully convolutional networks trained end-to-end for semantic segmentation achieve state-of-the-art results by adapting classification nets and using a skip architecture for detailed predictions.
Grégoire Montavon, Wojciech Samek, Klaus‐Robert Müller
This tutorial paper introduces deep neural network interpretability, focusing on layer-wise relevance propagation (LRP) for explaining predictions.
Arteum D. Bochevarov, Edward Harder, Thomas F. Hughes, et al.
Jaguar is a high-performance quantum chemistry software that uses pseudospectral approximation and parallelization for fast electronic structure predictions of medium to large molecular systems.
Fabian Gloeckle, Badr Youbi Idrissi, Baptiste Rozière, et al.
Training LLMs to predict multiple future tokens simultaneously improves sample efficiency, downstream performance, and inference speed.
Unknown
Selective Self-to-Supervised Fine-Tuning (S3FT) improves LLM fine-tuning by selectively using the model's own correct predictions or gold responses to reduce overfitting and boost generalization.