Thanks to insights from our ODSC West researchers, attendees, and instructors we’ve pulled together some of the trending machine learning topics of 2021. We’re excited to host some of the leading experts and top contributors in each of these topics. Here are a few of our top picks.
MLOps has been featured at ODSC since 2018, but it really took off in 2020. It seems every company is offering an MLOps solution and it’s easy to see why. As machine learning (ML) pipelines grow more complex, practitioners are embracing DevOps and data engineering best practices to build MLOps workflows and pipelines. This is especially true as static systems give way to adaptable machine learning systems that dynamically adjust to changes in data and other environmental factors. Kubernetes, kubeflow, and MLFlow are all trending platforms, and pipelines in frameworks like scikit-learn and TensorFlow (TFX) are also getting a lot of traction.
Machine learning topics such as continuous integration and continuous delivery of models are central to MLOps. More advanced topics such as continuous training, where a newly trained model is served at the end of the pipeline, and continuous production monitoring to ensure your models are performing in the wild are also trending. ODSC West will feature over 15 sessions on MLOPs and related topics.
Real-Time Machine Learning
Real-time machine learning models are a trending machine learning topic for 2021. More advanced machine learning operation (MLOps) and real-time data is a powerful combination, one that gives us real-time machine learning.
Traditionally, machine learning models are trained on batches of historical data. Training ML models by feeding them real-time live data to continuously improve the model is a marked improvement. Building an ML system that makes predictions in real-time will continue to trend in 2021 and beyond.
Meta-learning for Machine Learning
The search for more generalized machine learning models that can be trained for more than a single task continues when considering trending machine learning topics. Meta-learning, or ‘Learning to Learn’ allows machine learning algorithms to learn from other algorithms and combine these algorithms to build improved models.
In general, machine learning models “learn” patterns from the input features provided. In contrast, meta-learning models learn from the outputs and meta-data of other models that serve as inputs. In addition to machine learning, meta-learning can be applied to deep learning, reinforcement learning, and NLP. Why’s it trending? In addition to being a strong driver of AutoML, it helps address some of the bottlenecks of machine learning such as more accurate predictions, faster training (less data), and more generalized models.
Data Privacy with Federated Learning
The dual needs of vast amounts of private data to train ML algorithms and data privacy concerns continue to offer a conundrum for ML engineers. Federated learning offers one possible solution by allowing anonymous training. Various techniques are involved in anonymous training, such as homomorphic encryption and differential privacy. Secure Multiparty Computation (SMC) is perhaps one of the most exciting techniques. It allows multiple organizations to collaboratively train an agreed-upon algorithm without leaking private input data of that organization, ultimately producing a shared model. Expect more open source tools like TensorFlow Federated (TFF) to trend. If you are interested in learning more, ODSC West will host a number of talks on this topic.
NLP and NLU
NLP has been a strong focus area among other machine learning topics at ODSC West for a number of years and 2021 will be no exception. Fine-tuning pre-trained transformer models continue to offer quick returns on time invested and business applications abound. Advanced models like GPT-2 and GPT-3 that demonstrate the capability for transfer learning (learn a new task through the transfer of knowledge of a similar learned task) will be of interest. New fine-tuning techniques will continue to find interesting applications, such as adaptive fine-tuning, behavioral fine-tuning, and text-to-text fine-turning.
Cybersecurity Meets Machine Learning
The fields of cybersecurity, digital forensics, network security, and threat analysis can all use machine learning to boost their capabilities. Machine learning can help security teams sift through the enormous amount of data involved in these areas to better detect breaches, predict attacks, and forecast intrusion points. Automated fraud detection, threat modeling, and vulnerability modeling are just a few of the techniques that are now employing machine learning modeling.
On the other hand, there are also security risks associated with deep learning and machine learning models. These models are susceptible to adversarial attacks via misrepresentative or malicious data (white box attacks) or model extraction (black box attacks). Defending machine learning and deep learning systems against adversarial attacks is a growing area of concern. This is one of ODSC West’s newest focus areas among other machine learning topics.
Register for ODSC West 2021 to learn more about these Machine Learning Topics
ODSC West 2021 will be a more extensive event than ever before— featuring your choice of in-person sessions or virtual experiences available for everyone, from anywhere, to learn about these trending machine learning topics. Our summer sale for 60% off in-person and virtual passes ends Friday, so register now before prices go up.
Focus areas for ODSC West include Machine Learning | NLP | Deep Learning | Hands-on Training | Cybersecurity and ML | MLOps | Research Frontiers | Kickstart Bootcamp | Responsible AI | Big Data Analytics
Founder of ODSC and Software Architect specializing in, complex multi-platform systems across multiple industries including finance, healthcare, and education.