The Most Common Types of Generative AI
Generative Artificial intelligence (AI) is revolutionizing various fields, from creative art to drug discovery. With the development of generative AI models like GANs and RNNs, AI can create, analyze, and interpret content in new and exciting ways.
In creative art, GANs can generate new images, while style transfer can transfer styles between them. With image restoration, old or damaged images can be restored, and with neural doodle, new images can be created from sketches.
In marketing, chatbots and sentiment analysis can help businesses communicate with customers more efficiently, while recommendation systems and predictive analytics can provide insights into customer behavior and preferences.
In film making and VFX, RNNs can help with auto-editing, while deepfakes and face swapping can create realistic fake videos. With voice cloning, voices can be cloned, and with scene generation, new scenes can be created.
In fashion, GANs and style transfer can help generate new textile patterns, and recommender systems can provide personalized recommendations for customers. With body measurement, AI can measure body size for custom clothes.
In drug discovery, QSAR and DeepChem are used for machine learning-based drug discovery, while molecular generation and virtual screening can help identify potential drugs.
To get started with Gen AI in these fields, check out the most useful links for each use case in this graphic. Explore the power of Gen AI and see how it can take your work to the next level.
Creative Art
GANs: Generative Adversarial Networks https://github.com/NVlabs/stylegan2Style Transfer: Transfer style between images - https://github.com/jcjohnson/fast-neural-styleImage Restoration: Restore old or damaged images - https://github.com/jantic/DeOldifyDeepDream: Generate dream-like images - https://github.com/google/deepdreamNeural Doodle: Generate images from sketches - https://github.com/alexjc/neural-doodle
Marketing
Recommendation Systems - https://towardsdatascience.com/how-to-build-a-recommender-system-quickly-2f49c8e8ec8bChatbots: Conversational AI - https://rasa.com/Sentiment Analysis: Analyze emotions in text - https://github.com/cjhutto/vaderSentimentCustomer Segmentation: Group customers based on behavior - https://towardsdatascience.com/a-step-by-step-guide-to-customer-segmentation-in-python-6d11ccbdd1d6Predictive Analytics: Predict customer behavior - https://www.kaggle.com/shrutimechlearn/step-by-step-predictive-analytics-using-python
Film making/VFX
RNN: Recurrent Neural Networks - https://github.com/keras-team/keras/tree/master/examplesDeepfakes: Create realistic fake videos - https://github.com/deepfakes/faceswapFace Swapping: Swap faces in videos - https://github.com/deepfakes/faceswapVoice Cloning: Clone voices - https://github.com/CorentinJ/Real-Time-Voice-CloningScene Generation: Generate new scenes - https://github.com/NVlabs/SPADE
Fashion
GANs: Generative Adversarial Networks - https://github.com/eriklindernoren/Keras-GANStyle Transfer: Transfer style between images - https://github.com/lengstrom/fast-style-transferRecommender Systems - https://towardsdatascience.com/intro-to-recommendation-systems-and-how-to-build-simple-recommender-system-using-python-4e4b7db5c35dBody Measurement: Measure body size for custom clothes - https://www.sizemic.io/Textile Design: Generate new textile patterns - https://towardsdatascience.com/machine-learning-for-textiles-10fda2422e3b
Drug discovery
QSAR: Quantitative Structure-Activity Relationship - https://jcheminf.biomedcentral.com/articles/10.1186/s13321-018-0257-8DeepChem: Machine learning for drug discovery - https://deepchem.io/Molecular Generation: Generate new molecules - https://pubs.acs.org/doi/10.1021/acs.jcim.9b00647Protein Folding: Predict protein structure - https://deepmind.com/research/open-source/alphafoldVirtual Screening: Identify potential drugs - https://www.sciencedirect.com/science/article/abs/pii/S0960894X17304756