RealTime Odor Classification Through Sequential Bayesian Filtering http

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es Javier G. Monroy Javier Gonzalez-Jimenez

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es Toxic or dangerous chemicals Localization of multiple gas sources Gas distribution maps In real-time Absence of steady state ? Desirable: uncertainty about class prediction

Real-Time Odor Classification Through Sequential Bayesian Filtering Training: 4 classes Test: Only 1 class http: //mapir. isa. uma. es Only instantaneous e-nose readings are used as features.

Real-Time Odor Classification Through Sequential Bayesian Filtering SEGMENTATION • Complex. • Influences the classification performance. • Not real time. SLIDING WINDOW • Optimization for selecting the appropriate window size. http: //mapir. isa. uma. es

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es BAYES FILTER: Probabilistic framework to efficiently integrate information from previous e-nose observations. Bayesian network of the HMM for the real-time odor classification problem. Bel(Ct) = P(Ct|Z 1: t)

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es Bel(Ct) = P(Ct|Z 1: t) Bayes rules Observation likelihood Normalization constant + HMM conditional indep. assumptions Number of class labels Transition probability

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es Transition probability ps is the probability that at two consecutive instants of time the e-nose is exposed to the same odor ( that is, the same class). Observation likelihood We do not propose any new classifier. To apply SBF to the posterior provided by any probabilistic classifier working on the most recent e-nose observation: P(Ct|Zt)

Real-Time Odor Classification Through Sequential Bayesian Filtering Removing terms not dependent of P(Ct) + http: //mapir. isa. uma. es Marginal class probability P(Ct) is time-independent

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es EXPERIMENTAL RESULTS Real-time classification of volatiles in uncontrolled, real environments. Naive Bayes classifier working on the instantaneous response of the e-nose. No Mixture of gases, just Gas transitions! DATASETS DS-UMA: 4 -classes, array of 5 MOX gas sensors. DS-UCI: subset corresponding to the parameters L 4, Vh=5, fan=100, downsampled to 1 Hz, and restricted to only 4 gas-classes.

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es EXPERIMENTAL RESULTS For a 4 -class problem, ps=0. 25 indicates uniform class-transition probability. ps=0. 99 indicates that the most likely class at time «t» , is that at time «t-1»

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es EXPERIMENTAL RESULTS SBF is specially effective when the posterior probabilities have similar values, but not so much when one class has a much higher probability that the others (as is desired).

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es EXPERIMENTAL RESULTS WITH ARTIFICIALLY SIMULATED CHEMICAL TRANSITIONS

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es CONCLUSIONS Sequential Bayesian filtering (SBF) approach to the real-time classification of volatile substances. Integration of information from previous e-nose observations, without relying on segmentation or sliding window approaches. Evaluation with two public olfaction datasets (chemical transitions). Results show that SBF is particularly suitable for high dynamic environments where spurious class transitions produced by the sensor dynamics are effectively removed. FUTURE RESEARCH Test alternative classifiers (SVM). Comparison with sliding window techniques. Adaptive approaches to SBF to automatically adjust the parameters of the model (e. g. class transition probabilities).

Real-Time Odor Classification Through Sequential Bayesian Filtering http: //mapir. isa. uma. es THANK YOU!