如图B,是在t-1, t, t+1时刻network的参数传递(仅展示出forward-propagation中节点间相互decision情况)
2....v_t输入为None,recurrence从全零开始gibbs采样k次生成v_t,并计算相应u_t
注: line 156: (u_t, bv_t, bh_t), updates_train = theano.scan...(v_t, u_t), updates_generate = theano.scan(lambda u_tm1, *_: recurrence(None, u_tm1),outputs_info=[None...line 60: chain, updates = theano.scan(lambda v: gibbs_step(v)[1], outputs_info=[v], n_steps=k)的k表示Gibbs...4. line 156: (u_t, bv_t, bh_t), updates_train = theano.scan(lambda v_t, u_tm1, *_: recurrence(v_t, u_tm1