Honors Projects
Showing 161 - 170 of 662 Items
Date: 2025-01-01
Creator: Kavi Sarna
Access: Open access
- Squeeze-and-Excitation (SE) networks improve CNN feature learning by channel-wise attention, but their global pooling strategy discards spatial context. In this work, we reinterpret the SE block’s excitation mechanism as a convolution operation, which leads to a novel patched pooling design. Instead of global average pooling, we propose to divide feature maps into patches and pool within each patch, preserving local spatial information for attention. The excitation step is implemented with 1x1 convolutions (replacing the original SE fully-connected layers), enabling the model to learn adaptive channel reweighting efficiently across those patches. This Convolutional Squeeze-and-Excitation (CSE) approach yields spatially aware feature recalibration with minimal overhead. We evaluate CSE across multiple CNN architectures (including a custom ConvNet and ResNet) on image classification tasks (Fashion-MNIST, CIFAR-10). The results show consistent accuracy improvements over standard SE blocks. Moreover, we demonstrate the generality of patched pooling by integrating it with other attention modules like Efficient Channel Attention (ECA) and Global Context (GC), achieving further gains. Our findings highlight that incorporating localized pooling in SE-style attention significantly enhances representation learning across diverse scenarios.

Date: 2025-01-01
Creator: Luisa Isabelle Louchheim
Access: Access restricted to the Bowdoin Community

Date: 2025-01-01
Creator: Henry Grant Marriott
Access: Access restricted to the Bowdoin Community
Date: 2025-01-01
Creator: Isabelle Sungsil Lee
Access: Open access
Date: 2025-01-01
Creator: Maya Juliette Le
Access: Open access

Date: 2025-01-01
Creator: Ari Edward Bersch
Access: Access restricted to the Bowdoin Community

Date: 2025-01-01
Creator: Julia Smart
Access: Access restricted to the Bowdoin Community
Date: 2019-05-01
Creator: Anna Louisa Roosevelt Lennon
Access: Open access
Date: 2019-05-01
Creator: Sarena Sabine
Access: Open access
- Previous research has shown that experiences in nature are predictive of increased connection to nature, well-being, and pro-environmental behavior. The current study investigated whether daily exposure to indoor nature imagery would also improve well-being and eco-conscious behaviors, and whether personally relevant images would enhance this hypothesized effect. Participants completed a test assessing baseline connection to nature and well-being, specifically satisfaction with life, positive and negative emotions, and stress. In the 2 (Nature vs. Built) X 2 (Familiar vs. Unfamiliar) study design, 125 participants either received a poster from a photo that they submitted (a personally-relevant nature scene or personally-relevant built scene) or a poster of an unfamiliar natural or built scene. After four weeks of daily exposure to this new poster in their home, participants completed a post-test which included the same measures of well-being and connection to nature, along with a novel eco-conscious behaviors measure involving environmental petitions. The nature intervention significantly improved participants’ satisfaction with life. The personal relevancy of images did not enhance well-being, either alone nor in interaction with image content. The finding that daily exposure could lead to improved well-being has implications for addressing mental health concerns.

Date: 2019-01-01
Creator: Sadie LoGerfo-Olsen
Access: Access restricted to the Bowdoin Community