An efficient machine learning approach for activity recognition

The population of older people in western nations are growing drastically. Independent lifestyle is their preference and this leads to fall instances often. Falls kind results severe health issues or sometimes causes deadly damages the elderly people. Considering problem, it highly important come up with discovery systems. A machine learning framework has been proposed regard which covers both day movement recognition. Utilizing acceleration speed inputs from dual unrestricted repositories helpful diagnosing maximum seven movements. Acceleration angular velocity parameters can be used extract attributes that relevant time frequency domain. These then offered a classification algorithm. An attempt made check outcomes four different procedures for categorizing manual above-mentioned Artificial Neural Network (ANN), K-Nearest Neighbours (KNN), Quadratic Support Vector Machine (QSVM), FP growth. Power spectral density maximize classifier. data alone considered activity initial step. Experimental show KNN, ANN, QSVM, growth attain correctness 81.2%, 87.8%, 93.2%, 94.1%. accuracy detection touches 97.2% 99.1% no false positive values QSVM procedures. As next phase, taken autocorrelation function power obviously maximizes correctness. Projected adopted achieve notable improvement 85.8%, 91.8%, 96.1%, 97.7% accurateness 100% absence any incorrect alarm, finest attainable performance.