Raccolta temi desame Anno 201920

Pagina 1

1 NEUROENGINEERING 2019‐2020 EXAM SAMPLES QUESTIONS Question Considering the neural plasticity, we can assert that: 1. an artificial neuron may modify its function activation depending upon of the intensity of the inputs 2. I f t w o b i o l o g i c a l i n t e r c o n n e c t e d n e u r o n s a r e a c t i v e a t t h e s a m e time then the synaptic link in between is strengthened 3. a biological neuron can increase or decrease the number of acti ve axons depending on the synchronism of its inputs 4. an artificial neuron in a feed‐forward network can change its weights even after the training phase 5. in a biological neuron, the acti vation threshold undergoes adap tation depending upon the number of incoming spikes. Question The principle of the learning mechanism in a feed‐forward neural network with supervision rests on: 1. exploiting the coherence between interconnected neurons on either side of a synapse 2. minimizing the error between the predicted signal of the neurons in the output layer and the nominal output provided by the supervisor 3. minimizing the error between the predicted signal of the neurons in the hidden layer and the nominal output provided by the supervisor 4. the backpropagation of the signals of output layer up to the first layer 5. exploiting the synchronous activation ensured by the supervisor of all the interconnected neurons Question We would like to train a perceptron using the Delta rule. One can expect that: 1. The initial value of the weights does affect the final convergence 2. the neuron threshold has been already set up 3. the update of one weight is proportional to the value of the activation function of the neuron 4. the weight variation increases when I am closer to the optimal solution 5. as the amount of patterns rises the…