In communication systems, available bandwidth and required transmission bitrate are closely related. It is always needed to reduce data transmission bitrate to save the communication channel’s bandwidth. Coding process plays a vital role in saving the communication channel’s bandwidth by compressing the data and hence reducing the required transmission bitrate. Video data transmission consumes most of the available communication bandwidth nowadays, hence video coding is very important. Apart from the usual use case of capturing physical scenes, synthetic contents such as game streaming and computer screen recording have become common use cases of the video encoding tools that we have today. These types of videos call for special screen content coding (SCC) tools, which can take advantage of the underlying structures of the frame to further increase coding efficiency. Screen content video is a new type of video that typically shows a mix of computer-generated screen content blocks (SCBs) and natural content blocks (NCBs). High Efficiency Video Coding (HEVC) is mainly designed for NCBs while SCBs have different properties, hence creative coding tools for SCBs are required. On top of HEVC, the Screen Content Coding (SCC) extension was created to investigate new coding methods for screen content videos. For intra-prediction, SCC implements two extra coding modes intra block copy (IBC) mode and palette (PLT) mode. Consequently, the exhaustive mode searching significantly increases the computing cost of SCC. This research aims to suggest a solution for the high computational complexity of the encoder in the Screen Content Coding Extension in High Efficiency Video Coding (HEVC-SCC) standard and therefore reduces the encoding process time. As a result, two machine learning-based algorithms are proposed in this thesis to accelerate screen content encoding. Firstly, a new technique has been proposed to save encoding time while conserving coding efficiency. The proposed algorithm selects the suitable mode for each Coding Unit (CU) and skips unhelpful modes by two methods, which depend on skipping unwanted modes by Neural Network Classifiers. The first classifier is the Neural Network Classifier Based on Current Depth Features (NNC_CF). The second one is the Neural Network Classifier Based on Parent Depth Features (NNC_PF). The simulation results demonstrate the efficacy of the proposed technique. Afterwards, to increase the efficiency of CU classification, another technique is proposed. The second proposed technique depends on two classifiers to classify a CU as a NCB or SCB efficiently. The first one is a NN classifier, which has a different design compared to the proposed one in the first approach, and the other is an AdaBoost classifier (Ada_CL), which depends on a boosted decision stump algorithm. The two classifiers predict the CU type individually and the final decision for CU classification depends on both. The experimental results reveal that the suggested technique significantly decreases encoding time without sacrificing coding efficiency.