Browse Articles

In their paper, Kreiman et al. examined what it takes for an artificial neural network to be able to perform well on a new task without forgetting its previous knowledge. By comparing methods that stop task forgetting, they found that longer training times and maintenance of the most important connections in a particular task while training on a new one helped the neural network maintain its performance on both tasks. The authors hope that this proof-of-principle research will someday contribute to artificial intelligence that better mimics natural human intelligence.

The authors looked at which measurements from the NFL combine were the most predictive of success for defensemen in the NFL.

In this study, the authors use existing mathematical models to how high school athletes pace 800 m, 1600 m, and 3200 m distance track events compared to elite athletes.

This study investigates how bending radius, wavelength, and number of turns affect signal attenuation in single-mode optical fibers.

Here, the authors used surface electromyography to measure the effects of caffeine intake on the resting activity of muscles. They found a significant increase in the measured amplitude suggesting that caffeine intake increased the number of activated muscle fibers during rest. While previous research has focused on caffeine's effect on the contraction signals of muscles, this research suggests that its effects extend to even when a muscle is at rest.

The authors looked at whether retinal features, specifically lens clarity, optic nerve cupping, and hyperreflective foci, can serve as biomarkers for neuro-ophthalmological diseases, which share pathological mechanisms with neurodegeneration and may indicate broader neurological risk. Using multimodal clinical data and machine learning, they investigated the potential of retinal imaging as a complementary diagnostic tool for more accurate and accessible detection of neurological disorders.

Textile waste from the fashion industry is a major environmental pollutant, but recycling waste into novel building material is a strategy to reduce the negative effects. This manuscript characterized five different binders that can be used to repurpose textile waste into bricks for construction purposes. Water-based glue, cement, white cement, plaster of Paris, and epoxy resin were mixed with shredded textile waste, and the mechanical characteristics and thermal insulation of each brick type were measured. Bricks with increased mechanical strength had the poorest thermal resistance, and the contrasting properties would suit different building purposes. This work provides a first step in generating recycled textile bricks for construction in a circular economy framework.

This study investigates how the hyperparameters epochs and batch size affect the classification accuracy of a convolutional neural network (CNN) trained on pulsar candidate data. Our results reveal that accuracy improves with increasing number of epochs and smaller batch sizes, suggesting that with optimized hyperparameters, high accuracy may be achievable with minimal training. These findings offer insights that could help create more efficient machine learning classification models for pulsar signal detection, with the potential of accelerating pulsar discovery and advancing astrophysical research.

The authors looked at how the length of an audio clip used of a song impacted the ability to properly classify it by musical genre.

Robot-assisted minimally invasive surgery (RMIS) benefits from increased precision and faster recovery, with force feedback from the surgical tool being critical for control. Researchers tested the use of neural networks for detecting the vanishing point of the tool, a key element for force feedback.