4DEEPANALYTICS
Updated 901 days ago
Çekmeköy, Istanbul - Turkey, 34788
We are PhD team with Academic/Practical background on Deep Learning and other state of-the-art Machine Learning techniques, have experiences on Signal/Audio Analysis, Video Analytics, Image Processing and Neural Processing based problem solving...
We need at least 500 samples per classes of events for multi-classes (up to 5 classes) classification, but for classifying more than 5 classes the deep neural nets need big data that is why established data centers nowadays. We can handle the loss (error) minimization problem at training phases, generally the CNNs error will increase by increasing the number of classes if using amount of dataset not be increasing comparatively exponentially!