Browse Articles

Modern artificial neural networks have been remarkably successful in various applications, from speech recognition to computer vision. However, it remains less clear whether they can implement abstract concepts, which are essential to generalization and understanding. To address this problem, the authors investigated the above vs. below task, a simple concept-based task that honeybees can solve, using a conventional neural network. They found that networks achieved 100% test accuracy when a visual target was presented below a black bar, however only 50% test accuracy when a visual target was presented below a reference shape.

The authors looked at the ability subjects to rotation a 4D cube and how the ability to practice cube rotation impact their ability to understand 4D space.

Here the authors investigated the impact of player ability versus starting positions in the game of Nim under imperfect play, hypothesizing that player skill becomes the primary determinant of outcomes as pile sizes grow. Through computational simulations and a mathematical model, they demonstrated that favorable starting positions lose their advantage over time and provide key insights to help improve reinforcement learning algorithms in abstract, complex decision-making environments.

This experiment assessed the effectiveness of a virtual environment as a method of exposure in the treatment of high school students’ public speaking anxiety. The results show that participants’ heartbeat was higher when they wore a VR headset than when they did not.

This study used surveys, interviews, and XGBoost analysis to examine how experienced and novice entrepreneurs differ in their thinking and decision-making.

The authors studied the accuracy of two commonly used explainable AI techniques for detecting pneumonia in chest x-rays.