Handwritten Number Recognition
MNIST dataset is a dataset used to evaluate handwritten numeral classification problems.
Import Data import tensorflow
import keras
from keras.datasets import mnist
from matplotlib import pyplot as plt
import numpy as np
from keras.models import Sequential
from keras.layers import Dense
from keras.utils import np_utils
# import mnist dataset from Keras
(X_train,y_train),(X_validation,y_validation) = mnist.load_data()
# Display 4 Handwritten Digital Pictures

Scikit-Learn is a universal machine learning library with complete functions and provides helpful methods in depth learning models.
Keras class library provides a wrapper for the deep learning model. Keras deep learning model is packaged into a classification model or a regression model in Scikit-Learn, so that methods and functions in Scikit-Learn can be conveniently used.
KerasClassifier (for classification models)
KerasRegression (for regression model)
1, use cross-validation evaluation model

CNN and LSTM Realize Two Classifications of DNA Binding Proteins (python+keras Realization) Main Contents word to vector binding protein sequence modification word embedding CNN1D implementation LSTM implementation from __future__ import print_function import numpy as np import h5py from keras.models import model_from_json np.random.seed(1337) # for reproducibility from keras.preprocessing import sequence from keras.models import Sequential from keras.layers.core import Dense, Dropout, Activation from keras.

Vector,HashTable are thread-safe collection classes. However, these two classes are very early usage and should be used as little as possible now. set–No collection of duplicatesThere are three specific types of sets availableHashSet- based on the hash table set, the elements added to the hash table implement the hashCode () methodLinkedHashSet- returns elements in increasing order when iterating over a setTreeSet- Data Structure Based on

Keras Custom Loss Function Used in Scene Classification In the process of image scene classification, we need to customize the loss function and encounter many pits.Keras's own loss functions are all in the losses.py file.(The following defaults to classification processing) #losses.py #y_true是分类的标签，y_pred是分类中预测值（这

When building a neural network structure, we should add some network layers to the network. The following are some common network layers and their related usages.
1. Common Layer Common layers correspond to core modules. A series of common network layers are defined inside core, including full connection and activation layers.
1.Dense layer Dense Layer: Full Connection Layer.
keras.layers.core.Dense(output_dim, init='glorot_uniform', activation='linear', weights=None, W_regularizer=None, b_regularizer=None, activity_regularizer=None, W_constraint=None, b_constraint=None, bias=True, input_dim=None) output_dim: an integer greater than 0, representing the output dimension of the layer.

When training models, many things cannot be predicted at first.Especially you don't know how many rounds are needed to get the best verification loss.Usually, the simple method is to train enough rounds, at which time the model has already started fitting. According to the first run, determine the correct rounds required for training, and then start a new training from scratch with this optimal number of rounds.Of course, this method

Keras is a building block deep learning framework, which can be used to easily and intuitively build some common deep learning models.Before tensorflow came out, Keras was almost the most popular in-depth learning framework at that time, taking theano as the back end. Now Keras has supported four back ends at the same time: theano, tensorflow, cntk and mxnet (the first three official supports, mxnet has not yet been integrated

Experimental Report
0, BaseNet (Layer 3, sigmoid, 784-386-10)one
1, Hidden_Net(784-112-10 and 784-543-10)2.
2、reluActivation_Net 3
3, DeepNet (4, 5) 4
4, DeepNet (four, five layers;Increase in the number of training rounds) 5
5, DeepNet (five layers;Dropout） 6
7, DeepNet (five layers;Dropout+relu） 8
8, AutoEncoder_Net (five layers;AutoEncoder） 9
9, Conclusion 10
Abstract: The data set of this experiment is MNIST digits.