The IoT brings the power of the internet, data processing and analytics to the real world of physical objects. However, Attackers may attempt to exploit vulnerabilities in application protocols, including Domain Name System (DNS), Hyper Text Transfer Protocol (HTTP) and Message Queue Telemetry Transport (MQTT) that interact directly with backend database systems. Successful exploitation of the protocols can result in security breaches. Machine Learning is seen as a method to defend against malware, botnets and other attacks. Our goal is to classify the IOT traffic into attack and benign traffic by deploying a DNN which will train the model for features. These data can be tested upon to detect if there is any malicious activity or intrusion in the IOT network. This paper shows that how machine learning and deep learning techniques can be used to efficiently used to detect and classify malicious attacks on IOT devices. Thus by this way we will train our model for 2 labelled data sets namely attack traffic and benign traffic.