Autor: Saulo Tipas

Currently, we have several models ready to be used in different platforms and scenarios, and all these tools grow and improve according to needs and usage. Google's IA Platform, AWS IA Services, and Microsoft's Azure AI are the leading high-level software houses that offer not just 1 or 2 models for predictive analysis, but each provides a complete suite:

Google IA Platform:

At Google, we have access to both training and prediction platforms. In the training side, we can create our own models using TensorFlow, Scikit-learn, and XGBoost. We can also use pre-built algorithms with our dataset without writing a training application. The predictive part manages cloud processing resources to run models on a large scale and provides online and batch prediction requests.

AWS IA Services:

AWS integrates models for common use cases, such as creating personalized recommendations, as their APIs continuously learn. Although they claim that no knowledge of Machine Learning is required, validation and correction should be accompanied by a business expert and a technical person familiar with the models. Amazon Rekognition, Amazon Lookout for Vision, and AWS Panorama are part of the Artificial Vision set, which also includes automated data analysis and extraction, and language AI.

Now, let's focus on AWS's business metrics, where we have tools like Amazon Forecast, Amazon Fraud Detector, and Amazon Lookout for Metrics. These tools stand out due to their wide range of use cases and ease of implementation. The utility we provide to these available models can significantly help any organization in their inventory and workforce planning.

 

In Azure, we also find a plethora of AI-related possibilities integrated throughout the Azure suite, including its databases and the Power BI business intelligence tool. Automated ML, Generative AI, MLOps, and Responsible AI are some options, each accompanied by a catalog of available models ready to be tailored to the client's needs, without losing the option to include custom models.

Some clients may not have the complete suites of these platforms integrated, but this is no impediment to utilizing these models. Currently, there is a wide variety of open-source languages with robust libraries to design models with all the necessary specifications. Let's start with Meta's new release, LLAMA 2, which was recently launched, with its 3rd version expected soon. Another noteworthy tool is Falcon 180B, valued for its versatility in sentiment analysis, text classification, and language translation. We can also mention libraries like TensorFlow, OpenAI, PyTorch, Keras, and Caffe, among others, all robust and available for implementing AI or ML development.

As we can see, there is a wide range of tools and languages that have proven successful in their predictions and suggestions. The application of AI has shifted from being exclusive to being widespread across all business processes, improving decision accuracy in line with market trends.

¿What did you think? Share your opinion now!