Jonas Kubilius

This paper introduces a method for efficient, single-step genome editing of bacteriophages using targeted DNA nicking, achieving high recombination rates without reducing phage titers.

This study develops ConvRNN models to simulate primate ventral visual stream processing, finding that recurrent networks achieve better task performance and efficiency than deeper feedforward networks.

Among dozens of ImageNet-trained networks scored against macaque V4/IT recordings and human behavior, DenseNet-169, CORnet-S and ResNet-101 rank highest, and brain similarity stops tracking accuracy above 70% top-1.

The paper shows that recurrent circuits in the primate ventral stream are critical for object recognition, as challenge images require longer processing and are poorly predicted by feedforward models.

CORnet-S, a four-area recurrent network with areas pre-mapped to V1, V2, V4 and inferior temporal cortex, ranks among top Brain-Score models while having a shorter longest convolutional path than competitors.