Our Recent News
論文題目: "A 67μW/Channel, 0.13nW/Synapse/Bit Nose-on-a -Chip for Non-invasive Diagnosis of Diseases with On-chip Incremental Learning"
Posted by NBME on 10 11, 2024
論文題目: "Sensors-74881-2024.R1 A Dual-Path Deep Learning Model for Low-Cost Air Quality Sensor Calibration"
Posted by NBME on 09 13, 2024
論文題目: "A 0.9V Adaptive Sampling Rate Differential level-Crossing SAR ADC for Biomedical Signal Acquisition System"
Posted by NBME on 08 24, 2024
論文題目: "HAF: a High Pe Utilization Spike-Based CNN Accelerator with Adaptive Input Compression and Forward-Based Data Flow"
Posted by NBME on 08 10, 2024
論文題目: "A Concentration Separability Indicator (CSI) Feature Selection Method to Enhance Coffee Classification for an Electronic Nose System"
Posted by NBME on 07 20, 2024
論文題目: "Gas Identification Algorithm Based on Dynamic Response Analysis of Metal Oxide Sensors under Temperature Modulation"
Posted by NBME on 06 03, 2024
論文題目: "Implementation of a Double-Path Flipped Voltage Follower Cell for Voltage Conveyors"
Posted by NBME on 05 17, 2024
論文題目: "A 0.67-to-5.4 TSOPs/W Spiking Neural Network Accelerator with 128/256 Reconfigurable Neurons and Asynchronous Fully-Connected Synapses"
Posted by NBME on 05 07, 2024
Posted by NBME on 05 06, 2024
Posted by NBME on 05 02, 2024
Posted by NBME on 03 22, 2024
Posted by NBME on 02 19, 2024
Posted by NBME on 02 19, 2024
Posted by NBME on 01 16, 2024
Posted by NBME on 12 15, 2023
Posted by NBME on 11 23, 2023
Posted by NBME on 11 23, 2023
Posted by NBME on 10 26, 2023
Posted by NBME on 10 22, 2023
Posted by NBME on 10 22, 2023
Artificial Intelligence, Electronic Nose and Biomedical Implants
Recent Publications
Coffee drinks prepared from moldy coffee beans can adversely affect human health. No convenient screening method for detecting the smell of stale coffee beans exists. Accordingly, this study developed an electronic nose (E-nose) system for detecting the smell of coffee beans. This system comprises an environmental control system, a sensor array, and a data signal readout system. The system can distinguish various degrees of mold on coffee beans through the recognition of the smell of the coffee beans. In this study, we established a standard operating procedure to collect gas samples from coffee beans in a temperature- and humidity-controlled environment and recorded changes in the signals by using the sensor array after introducing the target gas. Features were first extracted from the collected data, then dimensionality reduction methods, such as principal component analysis and linear discriminant analysis, were applied to these features. Thus, their complexity was reduced, and the noise was eliminated. K-nearest neighbor and support vector machine were adopted as classification algorithms, and the classification accuracy of the proposed system reached 91.77%.
Non-maximum suppression (NMS) is an essential part of the post-processing of object detectors. Most object detection models require NMS algorithms to filter overlapping candidate boxes belonging to the same object and reserve the box that best represents the object as the object representative box. However, the hardware for the standard NMS algorithm has some disadvantages, such as high complexity (especially when box numbers are significant), high latency, large area, and high power consumption. To solve these problems, we propose an efficient parallel hardware architecture, which uses a new sorting circuit with a ping-pong buffer and a new retention mechanism of the candidate box for the new NMS algorithm called distance over side-NMS (DoS-NMS). This architecture uses a PE group with a voting mechanism to simplify the algorithm for reducing the latency and area. Additionally, the PE group computes the DoS (non-intersection over union) of the candidate box and multiple object representatives in parallel, significantly reducing algorithm complexity and memory access cost. Experiments indicated that the algorithm runs on the chip with area of 0.75mm2 , power consumption of 68.41 mW, and normalized area efficiency that is 3.72 and 4.84 times higher than the two state-of-the-art methods, respectively.
Contact Us