我编写了以下Keras神经网络代码，当我运行它时，准确率总是卡在0.4到0.6之间的值上。我选择的优化器是否有误，或者还有其他方法可以提高我的准确率？我有一个输入数组[8100:63]和一个输出数组[8100:3]。如果能得到一些帮助或建议，我将不胜感激。
这是我的代码：
输出总是看起来像这样：
100/8160 [..............................] - ETA: 0s - loss: 8.4386e-05 - acc: 0.60002100/8160 [======>.......................] - ETA: 0s - loss: 7.6640e-05 - acc: 0.56334000/8160 [============>................] - ETA: 0s - loss: 7.5545e-05 - acc: 0.56035600/8160 [===================>..........] - ETA: 0s - loss: 7.5711e-05 - acc: 0.55807300/8160 [=========================>....] - ETA: 0s - loss: 7.6259e-05 - acc: 0.55378160/8160 [==============================] - 0s 28us/step - loss: 7.6090e-05 - acc: 0.5522Epoch 497/500 100/8160 [..............................] - ETA: 0s - loss: 9.6210e-05 - acc: 0.59001600/8160 [====>.........................] - ETA: 0s - loss: 8.0017e-05 - acc: 0.55062900/8160 [=========>....................] - ETA: 0s - loss: 7.9372e-05 - acc: 0.55664300/8160 [==============>...............] - ETA: 0s - loss: 7.7604e-05 - acc: 0.55265900/8160 [====================>.........] - ETA: 0s - loss: 7.5976e-05 - acc: 0.55207600/8160 [==========================>...] - ETA: 0s - loss: 7.5226e-05 - acc: 0.54888160/8160 [==============================] - 0s 33us/step - loss: 7.5611e-05 - acc: 0.5515Epoch 498/500 100/8160 [..............................] - ETA: 0s - loss: 7.1056e-05 - acc: 0.54002000/8160 [======>.......................] - ETA: 0s - loss: 7.3529e-05 - acc: 0.53903900/8160 [============>................] - ETA: 0s - loss: 7.2863e-05 - acc: 0.55055800/8160 [====================>.........] - ETA: 0s - loss: 7.3346e-05 - acc: 0.55347200/8160 [=========================>....] - ETA: 0s - loss: 7.4003e-05 - acc: 0.55248160/8160 [==============================] - 0s 29us/step - loss: 7.4069e-05 - acc: 0.5522Epoch 499/500 100/8160 [..............................] - ETA: 0s - loss: 6.8331e-05 - acc: 0.53001900/8160 [=====>........................] - ETA: 0s - loss: 7.2856e-05 - acc: 0.54323800/8160 [============>.................] - ETA: 0s - loss: 7.3400e-05 - acc: 0.54245800/8160 [====================>.........] - ETA: 0s - loss: 7.4324e-05 - acc: 0.54917700/8160 [===========================>..] - ETA: 0s - loss: 7.5220e-05 - acc: 0.55318160/8160 [==============================] - 0s 27us/step - loss: 7.5057e-05 - acc: 0.5522Epoch 500/500 100/8160 [..............................] - ETA: 0s - loss: 7.8258e-05 - acc: 0.57002100/8160 [======>.......................] - ETA: 0s - loss: 8.3809e-05 - acc: 0.54954100/8160 [==============>...............] - ETA: 0s - loss: 8.1708e-05 - acc: 0.54346100/8160 [=====================>........] - ETA: 0s - loss: 7.9374e-05 - acc: 0.54757900/8160 [============================>.] - ETA: 0s - loss: 7.8028e-05 - acc: 0.54858160/8160 [==============================] - 0s 26us/step - loss: 7.7851e-05 - acc: 0.5478回答：
如果您在做回归任务，我建议将您关注的指标改为mse，以反映您训练的目标。
model.compile(optimizer=Optimizer, loss=['mse'], metrics=['mse'] )如果您在做分类任务（准确率是关于分类的），那么损失函数应该改为
model.compile(optimizer=Optimizer, loss=['categorical_crossentropy'], metrics=['acc'] )