Application of computational verb theory to gas recognition

Gas recognition techniques are important for metal oxide sensors in practical applications because of their poor selectivities. The conventional solution is using sensor arrays instead of single sensors for pattern recognition. But this solution is expensive and the accuracy is affected by every single sensor in it. So developing recognition techniques based on single sensors is strongly needed. In this paper, a new method based on computational verb theory was developed by employing verb similarities as decision features and a clear linear decision boundary was found. Precise recognition between ethanol and acetone was conducted by using a single sensor. Compared with the conventional gas recognition approach, this method is more cost effective and programmable which shows promising applications in future smart gas detecting systems.