Spiking Neural Networks (SNNs) represent the third generation of artificial neural networks, distinguished by their biomimetic nature and ability to process information similarly to the human brain. Unlike traditional neural networks, SNNs integrate dimensions like time, frequency, and phase using spike trains, mimicking neural coding in biological systems. This unique characteristic gives SNNs several advantages, including high computational speed, low power consumption, and biological interpretability.
The field of electronic noses (E-noses), which mimics the olfactory system for detecting and distinguishing odors, faces challenges in traditional signal processing methods. Feature extraction from sensor signals is often complex, requiring considerable expertise. Moreover, sensor signal drift can affect reproducibility, hampering the accuracy of E-nose systems. To address these issues, SNN-based processing can provide an efficient, automatic feature extraction approach that enhances odor discrimination and reduces the need for manual intervention. Traditional methods of feature extraction in E-noses often rely on manual, experience-based techniques, which can lead to inconsistent results and poor reproducibility. Spiking Neural Networks (SNNs) offer a promising alternative by automatically extracting spatiotemporal features from sensor data, similar to how the mammalian olfactory bulb processes odor signals.
Main tasks:
• Develop and implement an SNN-based odor discrimination model for electronic nose applications.
• Investigate spike coding strategies to convert sensor signals into meaningful spike sequences.
• Test the model’s performance in odor classification and compare it with traditional pattern recognition algorithms.
• Evaluate the system's ability to suppress sensor signal shifts and enhance feature extraction in real-time scenarios.
Student background:
• Good background in machine learning, neuroscience, signal processing, or artificial intelligence.
• Familiarity with neural networks and an understanding of biological signal processing is an advantage.
• Experience in programming (Python, MATLAB, or similar) and data analysis would be beneficial.
Benefits to the student:
• Gain hands-on experience with cutting-edge SNN algorithms and their application in real-world sensing technology.
• Learn about the intersection of machine learning, neuroscience, and sensor technology.
• Work on innovative projects with potential for publication and contribution to next-generation odor sensing systems.
• Develop expertise in a rapidly growing field with promising applications in environmental monitoring, healthcare, and industrial safety.
Reference:
[1]. Vanarse, Anup, et al. "Application of a brain-inspired spiking neural network architecture to odor data classification." Sensors 20.10 (2020): 2756.
[2]. Ambard, Maxime, et al. "A spiking neural network for gas discrimination using a tin oxide sensor array." 4th IEEE International Symposium on Electronic Design, Test and Applications (delta 2008). IEEE, 2008.
[3]. Yan, Jia, et al. "An SNN-based bionic olfactory signal processing network for odor recognition." IEEE Sensors Journal 23.12 (2023): 13186-13197.
Spiking Neural Networks (SNNs) represent the third generation of artificial neural networks, distinguished by their biomimetic nature and ability to process information similarly to the human brain. Unlike traditional neural networks, SNNs integrate dimensions like time, frequency, and phase using spike trains, mimicking neural coding in biological systems. This unique characteristic gives SNNs several advantages, including high computational speed, low power consumption, and biological interpretability.
The field of electronic noses (E-noses), which mimics the olfactory system for detecting and distinguishing odors, faces challenges in traditional signal processing methods. Feature extraction from sensor signals is often complex, requiring considerable expertise. Moreover, sensor signal drift can affect reproducibility, hampering the accuracy of E-nose systems. To address these issues, SNN-based processing can provide an efficient, automatic feature extraction approach that enhances odor discrimination and reduces the need for manual intervention. Traditional methods of feature extraction in E-noses often rely on manual, experience-based techniques, which can lead to inconsistent results and poor reproducibility. Spiking Neural Networks (SNNs) offer a promising alternative by automatically extracting spatiotemporal features from sensor data, similar to how the mammalian olfactory bulb processes odor signals.
Main tasks:
• Develop and implement an SNN-based odor discrimination model for electronic nose applications.
• Investigate spike coding strategies to convert sensor signals into meaningful spike sequences.
• Test the model’s performance in odor classification and compare it with traditional pattern recognition algorithms.
• Evaluate the system's ability to suppress sensor signal shifts and enhance feature extraction in real-time scenarios.
Student background:
• Good background in machine learning, neuroscience, signal processing, or artificial intelligence.
• Familiarity with neural networks and an understanding of biological signal processing is an advantage.
• Experience in programming (Python, MATLAB, or similar) and data analysis would be beneficial.
Benefits to the student:
• Gain hands-on experience with cutting-edge SNN algorithms and their application in real-world sensing technology.
• Learn about the intersection of machine learning, neuroscience, and sensor technology.
• Work on innovative projects with potential for publication and contribution to next-generation odor sensing systems.
• Develop expertise in a rapidly growing field with promising applications in environmental monitoring, healthcare, and industrial safety.
Reference:
[1]. Vanarse, Anup, et al. "Application of a brain-inspired spiking neural network architecture to odor data classification." Sensors 20.10 (2020): 2756.
[2]. Ambard, Maxime, et al. "A spiking neural network for gas discrimination using a tin oxide sensor array." 4th IEEE International Symposium on Electronic Design, Test and Applications (delta 2008). IEEE, 2008.
[3]. Yan, Jia, et al. "An SNN-based bionic olfactory signal processing network for odor recognition." IEEE Sensors Journal 23.12 (2023): 13186-13197.