def tall_loss(self, positive, negative, query, lw=1):
scores_p = self.distance_function(positive, query)
scores_n = self.distance_function(negative, query)
alpha_c = 1
alpha_w = 1
exp_p = L.Exp(scores_p, scale=-1)
exp_n = L.Exp(scores_n)
log_p = L.Log(exp_p, shift=1)
log_n = L.Log(exp_n, shift=1)
scale_p = L.Power(log_p, scale=alpha_c)
scale_n = L.Power(log_n, scale=alpha_w)
all_scores = L.Concat(scale_p, scale_n, axis=0)
return L.Reduction(all_scores, operation=4, loss_weight=[lw])
def weight_edges(bottom):
bottom_avg = L.Convolution(bottom,
convolution_param=dict(num_output=9,
kernel_size=1,
stride=1,
pad=0,
bias_term=False,
weight_filler=dict(
type='constant',
value=1.0)),
param=[{
'lr_mult': 0,
'decay_mult': 0
}])
weight = L.Exp(bottom_avg, exp_param=dict(scale=-1.0))
return weight
def test_exp4(self):
n = caffe.NetSpec()
n.input1 = L.Input(shape=make_shape([6, 4, 64, 64]))
n.exp1 = L.Exp(n.input1, base=2.0, scale=0.5)
self._test_model(*self._netspec_to_model(n, 'exp4'))
def test_exp3(self):
n = caffe.NetSpec()
n.input1 = L.Input(shape=make_shape([6, 4, 64, 64]))
n.exp1 = L.Exp(n.input1, scale=0.5, shift=0.01)
self._test_model(*self._netspec_to_model(n, 'exp3'))
def net():
n = caffe.NetSpec()
n.data = L.Input(input_param=dict(shape=dict(dim=data_shape)))
n.dataout = L.Exp(n.data, base=_base, scale=_scale, shift=_shift)
return n.to_proto()
