Science

Evolving Phylogenies with Deep Learning

by

A six-layer ELU network, trained on JC alignments with 779 parameters, effectively approximates the true Jukes-Cantor distance function - specifically, $4\ln(1-3x/4)/3$ - exhibiting appropriate behavior by predicting a constant value beyond $x=0.75$, where the distance diverges, and even surpassing the performance of a 50-term Maclaurin series approximation, suggesting the learned transformations capture implicit information about the underlying evolutionary process-as evidenced by a ceiling value of 4.840 exceeding the mean tree diameter (3.697) of the training dataset.

New research demonstrates how artificial neural networks can learn effective distance metrics from sequence data, potentially accelerating and improving the accuracy of evolutionary relationship inference.

Can AI Truly Innovate?

by

A new benchmark reveals that while artificial intelligence agents can generate novel solutions, they often do so at the expense of reliable performance.

The AI Composer in the Mind

by

Human perception of authenticity varies significantly by genre, with stimulus characteristics playing a crucial role in determining perceived humanness.

New research suggests our opinions of artificial intelligence, not the music itself, largely shape how we perceive and enjoy AI-generated compositions.