AI Trends
Reporting, analysis and features from AIM's AI Trends section.
How to ensemble neural networks using AdaNet?
In machine learning, ensemble approaches combine many weak learners to achieve better prediction performance than each of the constituent learning algorithms alone.
Vijaysinh Lendave ·
Accelerate training with multiple GPUs using PyTorch Lightning
PyTorch Lightning enables the usage of multiple GPUs to accelerate the training process. It uses various stratergies accordingly to accelerate training process.
darshan.m ·
A deep dive into image data preprocessing by TensorFlow
Preprocessing prevents from overfitting
Sourabh Mehta ·
Create interactive presentations within the Python notebook with ipyVizzu
The "ipyvizzu" is an animated graphing tool for notebooks
Sourabh Mehta ·
Building a causal inference model for medical analysis using DoWhy
Microsoft has formulated a library named DoWhy to build efficient and robust causal models for critical domains.
darshan.m ·
Neural vocoder and its application in speech recognition
Vocoders based on deep neural networks may produce human-like voices
Sourabh Mehta ·
TorchRec Sharding - A framework to build large-scale recommender systems
Pytorch has formulated a framework named TorchRec sharding that is suitable for training and modelling large scale recommendation systems.
darshan.m ·
A new programming language that might be successor to C++
Carbon reduces existing libraries' tech debt and makes extending those libraries easier.
Pritam Bordoloi ·
Building robust models with learning rate schedulers in PyTorch?
Pytorch learning rate scheduler is used to find the optimal learning rate for various models by conisdering the model architecture and parameters.
darshan.m ·
How to automate finding the optimal learning rate?
The Cyclic Learning rate method finds the rate automatically.
Sourabh Mehta ·
Bonsai Brain - A low code platform to build AI agents
The Bonsai Brain is a low code AI component that is integrated with Automation systems. The Bonsai Brain focuses on adding value to various Autonomous and AI systems.
darshan.m ·
Why should the learning rate always be low?
Learning rate controls the amount of information to be learned.
Sourabh Mehta ·
The Power & Pitfalls of AI in Indian Justice system
AI can help the judiciary dispose of thousands of pending cases.
Pritam Bordoloi ·
Best MLOps workflow to upscale ML lifecycles
MLops streamlines the process of production, maintaining and monitoring the ML model
Sourabh Mehta ·
Plug-and-play ML models across platforms using MLCube
MLCube is an interface which facilitates easy model sharing across platforms and account for increased model usage across platforms.
darshan.m ·
Structural time-series modelling with TensorFlow Probability
The observed data is forecasted using Bayesian Neural Networks
Sourabh Mehta ·
How evolutionary algorithms can optimize sorting?
Sorting and selection primitives acts an nodes in evolutionary tree algorithm
Sourabh Mehta ·
Tensor2Tensor to accelerate training of complex machine learning models
This article mainly focuses on the Tensor2Tensor library and to understand the dynamic abilities to handle and process complex models.
darshan.m ·
How to accelerate TensorFlow models with the XLA compiler?
XLA is a compiler used to accelerate training time of Tensorflow models and reduce memory consumption.
darshan.m ·
Tensorflow Model Remediation- A framework for responsible AI
Tensorflow model remediation is a framework used to obtain fairness free and bias free models. It aims to produce robust models that is not affected by sensitive attributes of data.
darshan.m ·