ML.NET: Machine Learning for .NET Developers

Machine Learning in .Net

ML.NET is a free software machine learning library for the C# and F# programming languages. It also supports Python models when used together with NimbusML. The preview release of ML.NET included transforms for feature engineering like n-gram creation, and learners to handle binary classification, multi-class classification, and regression tasks. Additional ML tasks like anomaly detection and recommendation systems have since been added, and other approaches like deep learning will be included in future versions.

ML.NET brings model-based Machine Learning analytic and prediction capabilities to existing .NET developers. The framework is built upon .NET Core and .NET Standard inheriting the ability to run cross-platform on Linux, Windows and macOS.

Developers can train a Machine Learning Model or reuse an existing Model by a 3rd party and run it on any environment offline. This means developers do not need to have a background in Data Science to use the framework. 

Build for .Net Developers

    With ML.Net, you can use your existing skills to easily integrate ML into your existing .Net applications without any prior experience.

ML.Net Performance

    Microsoft's paper on machine learning with ML.NET demonstrated it is capable of training sentiment analysis models using large datasets while achieving high accuracy. Its results showed 95% accuracy on Amazon's 9GB review dataset.

NimbusML Python support

    Microsoft acknowledged that the Python programming language is popular with Data Scientists, so it has introduced NimbusML the experimental Python bindings for ML.NET. This enables users to train and use machine learning models in Python. It was made open source similar to Infer.NET.

ML.Net capabilities

• Sentiment Analysis - Analyse sentiment of customer review using binary classification algorithm.
• Product recommendation - Recommends products based on purchase history using a matrix factorization algorithm.
• Price prediction - Predict taxi fare based on parameters such as distance traveled using regression algorithm.
• Customer segmentation - Identify group of customers with similar profiles using clustering algorithm.
• Object detection - Recognize objects in an image using an ONNX deep learning model.
• Fraud detection - Detect fraudulent credit card transaction using a binary classification algorithm.
• Sales spike detection -  Detect spikes and changes in product sales using an anomaly detection model.
• Image classification - Classify images using TensorFlow deep learning model.
• Sales forecasting - Forecast further sales for product using a regression algorithm. 

Reference: ML.Net, wiki

Web API Token Based Authentication

ASP.NET Web API can be accessed over Http by any client using the Http protocol. This framework enables data communication in JSON format (by default) and hence helps in lightweight communication.

Token based authentication

Since the Web API adoption is increasing at a rapid pace, there is a serious need for implementing security for all types of clients trying to access data from Web API services. One of the most preferred mechanism is to authenticate client over HTTP using a signed token. Simply put, a token is a piece of data which is created by a server, and which contains enough data to identify a particular user. The process starts by allowing users to enter their username and password which accessing a service. Once the user provides the username/password, a token is issued which allows users to fetch a specific resource - without using their username and password every time. This token is sent to the server with each request made by the client and contains all necessary information to validate a user’s request. The following diagram explains how Token-Based authentication is used in communication between clients and server.
Token based authentication works by ensuring that each request to a server is accompanied by a signed token which the server verifies for authenticity and only then responds to the request.

Why use Tokens based authentication?

1. Tokens are stateless - The token is self-contained and contains all the information it needs for authentication. This is great for scalability as it frees your server from having to store session state.
2. Tokens can be generated from anywhere -  Token generation is decoupled from token verification allowing you the option to handle the signing of tokens on a separate server or even through a different company such us Auth0. 
3. Fine grained access control - Within the token payload you can easily specify user roles and permissions as well as resources that the user can access.
4. Mobile Friendly - This type of authentication does not require cookies, so this authentication type can be used with mobile applications.
5. Loosely Coupling- Your front-end application is not coupled with specific authentication mechanism, the token is generated from the server and your API is built in a way to understand this token and do the authentication.

JSON Web Token (JWT) is an open standard (RFC 7519) that defines a compact and self-contained method for securely transmitting information between parties encoded as a JSON object. JWT has gained mass popularity due to its compact size which allows tokens to be easily transmitted via query strings, header attributes and within the body of a POST request.

A JSON Web Token consists of three parts: Header, Payload and Signature. The header and payload are Base64 encoded, then concatenated by a period, finally the result is algorithmically signed producing a token in the form of header.claims.signature. The header consists of metadata including the type of token and the hashing algorithm used to sign the token. The payload contains the claims data that the token is encoding.

JSON Web Token Best practices

1. Keep it secret. Keep it safe. The signing key should be treated like any other credentials and revealed only to services that absolutely need it.
2. Do not add sensitive data to the payload. Tokens are signed to protect against manipulation and are easily decoded. Add the bare minimum number of claims to the payload for best performance and security.
3. Give tokens an expiration. Technically, once a token is signed – it is valid forever – unless the signing key is changed or expiration explicitly set. This could pose potential issues so have a strategy for expiring and/or revoking tokens.

4. Embrace HTTPS. Do not send tokens over non-HTTPS connections as those requests can be intercepted and tokens compromised.

World of Constructors in C#

Constructors

     Constructor are class methods that are executed automatically when instance of class is created. Constructors are use to initialize globally declared members of given class. Constructors can only run once when instance of class is created or in other words memory is allocated for given class and even it constructor runs before any written code in class. We can have multiple constructors in class, with help of polymorphism in way of overloading. If we do not define any constructor in class then compiler will automatically create Default constructor in the class.

Notable points about Constructors

üA Class can have any numbers of constructors.

üA static constructor does not have any parameters.

üWith in class you can create only one static constructors.

üA constructor does not have any return type, not even void.

Types of Constructors

1. Default constructor - A Constructor without any parameter is called a default constructor. In this constructor every instance of class will be initialized without any parameters values. The default constructor initializes all numeric fields to zero and all strings and object to null.

2. Parametrized constructor - A Constructor with at least one parameter is called a parameterized constructor. This good side of this constructor is that you can initialize each instance with different values.

a) Constructor Overloading - We can overload constructor by creating another constructor with same method name with different parameters.

3. Copy constructor - A Parameterized constructor that contains a parameter of same class type is called copy constructor. The main purpose of copy constructor is to initialize new instance to the value of an existing instance.

4. Static constructor - When we declare constructor as static it will be invoked only once for any number of instances of class and its during creation of first instance or first reference to a static member in the class. Static constructor is used to initialize static fields of the class and to write the code that needs to be executed only once.

a) Notable points of Static constructor

i. A static constructor does not take access modifiers or have parameters.

ii. A static constructor cannot be called directly.

iii. The user has not control on when the static constructor is executed in the program because it is called by CLR during compile time.

iv. In Static constructors constants variable can be created, where values does not changes after initialization.

5. Private constructor - Private constructor is a special instance constructor used in a class that contains static member only. If a class has private constructor and no public constructor then other classes is not allowed to create instance of this class this mean we can neither create the object of the class nor it can be inherit by other class. The main purpose of creating private constructor is used to restrict the class from being instantiated when it contains every member as static.

6.  Instance constructor - A Constructor without any parameter is called instance constructor. This sounds similar to Default constructor and static constructor, but it is different.

a) Default constructor initializes default values of variable as per data types, by in instance constructor we can assign specific values to variables.

              b) In static constructor, it get called automatically from CLR, but instance constructor get called when instance of class is created.

Destructor

     Destructor are use to release memory allocated for instance of object. In .Net framework, the garbage collector automatically manages the allocation and release of memory for manages objects in your application. But there were still unmanaged code, which can be free up specifically by using Destructor.

Difference between Object Vs Var Vs Dynamic Keyword

Object Vs Var Vs Dynamic Keyword

So right now in C# we have the Object class, and the var and dynamic types. At first look, they all seem to do the same job, but not really.

Then the question is, what is the difference between Object, var and dynamic? And when should we use them?

So here is the answer.

Object

The object class in C# represents the System.Object type, which is the root type in the C# class hierarchy. Generally we use this class when we cannot specify the object type at compile time, which generally happens, when we deal with interoperability.

Let's have an example. The following example explains that the variable amount, of which the type is object, but at run time we can get the actual type of that variable that is stored in the variable.




Let's perform a mathematical operation on it.



Why are we unable to perform a mathematical operation on it and instead get an error message but in the previous example we get the type of Amount as System.Int32. If the amount is a Systme.Int32 type and we can store an integer value in it then why are we unable to apply a simple mathematical operation?

Here is the reason. Actually, an object requires explicit type conversion before it can be used. We can store anything in the object type variable but for performing an operation we must type cast it. Because C# is a statically typed language so it will throw an exception when we start performing any operation on it without proper type casting.



Var

The var type was introduced in C# 3.0. It is used for implicitly typed local variables and for anonymous types. The var keyword is generally used with LINQ.

When we declare a variable as a var type, the variable's type is inferred from the initialization string at compile time.



We cannot change the type of these variables at runtime. If the compiler can't infer the type, it produces a compilation error.



Dynamic

The dynamic type was introduced in C# 4.0. The dynamic type uses System.Object indirectly but it does not require explicit type casting for any operation at runtime, because it identifies the types at runtime only.



In the code above we are assigning various types of values in the variable amount because its type is dynamic and dynamic delays determination of the type until execution. All dynamic types variables enjoy the party at runtime.

How to use Datasets with Linq in Asp.net

Define:

             Linq to Datasets provides a runtime infrastructure for querying Datable in Datasets as object without losing ability to querying. Linq to Data ships in .Net Framework 4.0.  Lin treats dataset as a Collection of columns which hold relational data. Let’s start with Simple Linq to dataset query and further in deep functionality of query.

Data:

First we will be having data in two data tables which resides in function GetEmployee() and GetDepartment(). Below is brief look at two data tables and structure.

Employee Table

Department Table

These datatables can be further added to datasets in the following ways.

DataSet ds = new DataSet();

       ds.Tables.Add(getEmployee());

       ds.Tables[0].TableName = "Emp";

       ds.Tables.Add(getDapartment());

       ds.Tables[1].TableName = "Dept";

We can even define Foreign key relation between two tables in dataset.

                DataColumn[] dc;

       dc = new DataColumn[1];

dc[0] = ds.Tables["Dept"].Columns["DeptID"];

       ds.Tables["Dept"].PrimaryKey = dc;

ForeignKeyConstraint fkcDept = new ForeignKeyConstraint(ds.Tables["Dept"].Columns["DeptID"],

ds.Tables["Emp"].Columns["DeptID"]);

       ds.Tables["Emp"].Constraints.Add(fkcDept);

Select Linq Query on Dataset

Employee Table

var query = (from a in ds.Tables["Emp"].AsEnumerable()

       select new

       {

EmpId = a["EmpId"],

              Name = a["Name"],

              Salary = a["Salary"],

});

gvData.DataSource = query;

       gvData.DataBind();

Result

Department Table

var query = (from a in ds.Tables["Dept"].AsEnumerable()

       select new

       {

DeptId = a["DeptId"],

              Name = a["DeptName"],

});

gvData.DataSource = query;

       gvData.DataBind();

Result

Select Join Linq Query on Dataset

var query = (from a in ds.Tables["Emp"].AsEnumerable()

join b in ds.Tables["Dept"].AsEnumerable()

on a.Field<Int32>("DeptId") equals b.Field<Int32>("DeptId")

              select new

              {

                     EmpId = a["EmpId"],

                     Name = a["Name"],

                     DeptName = b["DeptName"],

                     Salary = a["Salary"],

});

Result

Where Linq Query on Dataset

var query = (from a in ds.Tables["Emp"].AsEnumerable()

join b in ds.Tables["Dept"].AsEnumerable()

on a.Field<Int32>("DeptId") equals b.Field<Int32>("DeptId")

       where b.Field<String>("DeptName") == "Information Technology"

       && a.Field<Decimal>("Salary") > 15000

       select new

       {

              EmpId = a["EmpId"],

              Name = a["Name"],

              DeptName = b["DeptName"],

              Salary = a["Salary"],

});

Result