Eating behaviors are developed gradually during childhood, so it is crucial to educate students about the nutrition of snacks and drinks sold in the market from an early age. This underscores the importance of school education in shaping and cultivating children's eating habits. Numerous studies have shown that taking students to supermarkets for nutrition education can improve their awareness of nutrition labels and buying habits. However, most nutritional education courses in Taiwan's elementary schools are conducted in traditional classrooms and do not effectively incorporate real-world learning situations. To improve the effectiveness of nutrition education in elementary schools, this study developed an AR food education system that combined game design, augmented reality, and image recognition technologies to support nutrition teaching activities. An experiment was conducted in an elementary school, inviting 27 6th grade students to participate and evaluate the effect of the proposed system on nutrition education. Results showed that the system attracted students' attention and provided a sense of satisfaction after learning. Moreover, the proposed approach enhanced students' learning performance in nutrition education.

Manipulating 3D objects has been among the active research topic for 3D vision. With the development and success of neural radiance field (NeRF) [1] on scene modeling, synthesizing and manipulating 3D objects using such a representation becomes desirable. In this paper, we introduce a semantic-aware generative NeRF, which is able to interpret the latent representation learned by category-specific generative NeRFs and to achieve editing of particular part attributes. With pretrained generative NeRF, we propose to deploy a semantic segmentor for performing part segmentation on the object category. This allows the rendering of the 2D image and prediction of the corresponding segmentation mask. Our proposed scheme learns to manipulate the resulting latent representation, optimized to edit the object part of interest with varying degrees. We conduct experiments on various object categories on benchmark datasets, and the results successfully verify the effectiveness and practicality of our proposed model.

Circuit learning has gained significant attention due to machine learning advancements and approximate synthesis applications. The task is to learn a circuit to model an unknown Boolean function subject to different design constraints. When circuit size is hard constrained, decision-tree-based learning plays a crucial role in state-of-the-art methods. However, it can be ineffective due to its structural restriction. This work proposes graph learning to overcome the limitation, provide tradeoffs between circuit size and accuracy, and enrich the portfolio of circuit learning tools. Experimental results show the superiority of our approach to prior work in accuracy, training time, and circuit size.

Hardware acceleration enables neural network (NN) inferencing on edge devices and for high throughput applications. Most approaches use neural processing elements for computation while storing weights in memory blocks. To avoid costly memory access, recent efforts seek direct logic implementation with weights hardwired into the circuit. However, special training strategies are often needed, and they could not maintain accuracy. In contrast, we take a trained and quantized NN as input and synthesize it by Booth encoding and logic sharing, resulting in a hardware accelerator without degrading accuracy. Experiments demonstrate that our method outperforms existing work in area reduction and/or throughput and power efficiency.

Logic synthesis is a fundamental step in hardware design whose goal is to find structural representations of Boolean functions while minimizing delay and area. If the function is completely-specified, the implementation accurately represents the function. If the function is incompletely-specified, the implementation has to be true only on the care set. While most of the algorithms in logic synthesis rely on SAT and Boolean methods to exactly implement the care set, we investigate learning in logic synthesis, attempting to trade exactness for generalization. This work is directly related to machine learning where the care set is the training set and the implementation is expected to generalize on a validation set. We present learning incompletely-specified functions based on the results of a competition conducted at IWLS 2020. The goal of the competition was to implement 100 functions given by a set of care minterms for training, while testing the implementation using a set of validation minterms sampled from the same function. We make this benchmark suite available and offer a detailed comparative analysis of the different approaches to learning.