Search for: All records

Why do brains have inhibitory connections? Why do deep networks have negative weights? We propose an answer from the perspective of representation capacity. We believe representing functions is the primary role of both (i) the brain in natural intelligence, and (ii) deep networks in artificial intelligence. Our answer to why there are inhibitory/negative weights is: to learn more functions. We prove that, in the absence of negative weights, neural networks with non-decreasing activation functions are not universal approximators. While this may be an intuitive result to some, to the best of our knowledge, there is no formal theory, in either machine learning or neuroscience, that demonstrates why negative weights are crucial in the context of representation capacity. Further, we provide insights on the geometric properties of the representation space that non-negative deep networks cannot represent. We expect these insights will yield a deeper understanding of more sophisticated inductive priors imposed on the distribution of weights that lead to more efficient biological and machine learning. 

Natural intelligences (NIs) thrive in a dynamic world – they learn quickly, sometimes with only a few samples. In contrast, artificial intelligences (AIs) typically learn with a prohibitive number of training samples and computational power. What design principle difference between NI and AI could contribute to such a discrepancy? Here, we investigate the role of weight polarity: development processes initialize NIs with advantageous polarity configurations; as NIs grow and learn, synapse magnitudes update, yet polarities are largely kept unchanged. We demonstrate with simulation and image classification tasks that if weight polarities are adequately set a priori, then networks learn with less time and data. We also explicitly illustrate situations in which a priori setting the weight polarities is disadvantageous for networks. Our work illustrates the value of weight polarities from the perspective of statistical and computational efficiency during learning. 

Most physical therapists would agree that physical rehabilitation is difficult to perform remotely. Consequently, the global COVID-19 pandemic has forced many physical therapists and their clients to adapt to telehealth, especially with video conferencing. In this article, we ask: How has telehealth for physical rehabilitation evolved with the global pandemic and what are the largest technological needs, treatment methodologies, and patient barriers? With the increased widespread use of telehealth for physical therapy, we present a qualitative study towards examining the shortcomings of current physical therapy mediums and how to steer future virtual reality technologies to promote remote patient evaluation and rehabilitation. We interviewed 130 physical rehabilitation professionals across the United States through video conferencing during the COVID19 pandemic from July—August 2020. Interviews lasted 30–45 min using a semi-structured template developed from an initial pilot of 20 interviews to examine potential barriers, facilitators, and technological needs. Our findings suggest that physical therapists utilizing existing telehealth solutions have lost their ability to feel their patients’ injuries, easily assess range of motion and strength, and freely move about to examine their movements when using telehealth. This makes it difficult to fully evaluate a patient and many feel that they are more of a “life coach” giving advice to a patient rather than a traditional in-person rehabilitation session. The most common solutions that emerged during the interviews include: immersive technologies which allow physical therapists and clients 1) to remotely walk around each other in 3D, 2) enable evidence-based measures, 3) automate documentation, and 4) provider clinical practice operation through the cloud. We conclude with a discussion on opportunities for immersive virtual reality towards telehealth for physical rehabilitation.