DeepMind wants to reconcile Deep Learning and classical computer science algorithms with Neural Algorithmic Reasoning. Featuring DeepMind's Petar Veličković and Charles Blundell, MILA's Andreea Deac
- S2E28
- 58:14
- September 10th 2021
Will Deep Learning really be able to do everything? We don't really know.
But if it's going to, it will have to assimilate how classical computer science algorithms work. This is what DeepMind is working on, and its success is important to the eventual uptake of neural networks in wider commercial applications.
This work goes by the name of Neural Algorithmic Reasoning. Join us as we discuss roots and first principles, the defining characteristics, similarities and differences of algorithms and Deep Learning models with the people who came up with this.
We also cover the details of how Neural Algorithmic Reasoning works, as well as future directions and applications in areas such as path finding for Google Maps
Could this be the one algorithm to rule them all?
Article published on VentureBeat.
Image: Getty
Orchestrate all the Things
Connecting the dots with George Anadiotis: Analyst, Consultant, Engineer, Founder, Host, Researcher, and Writer.
Stories about Tech, Data, AI and Media, and how they flow into each other shaping our lives.
I’ve engaged from the likes of Gary Marcus and Andrew Ng to emerging thinkers and innovators across multiple domains.
My stories have been featured on ZDNet and VentureBeat, and are syndicated across DZone, Hackernoon, Medium and Substack.
Some might call this futurism; let’s just say it’s connecting the dots
Many conversations have a technical focus. Most also examine business perspectives and use cases, while others are socio-technical.
Some are analyses on emerging themes – picking them up early, featuring expert comment, or offering alternative takes.
Others cover breaking news, typically also featuring the people behind them plus some analysis. There are some book reviews as well.
I focus on the connection between data, analytics, data science, graphs, machine learning and AI and their impact on society and business.
I have been covering topics related to:
- AI and Machine Learning
- Data, Analytics and Data Science
- Knowledge Graphs, Graph Databases, Graph AI & Data Science
- Innovation, and a wide array of technologies such as Blockchain, Cloud, Observability, IoT, Open Data and Open Source, Social Media and Software Engineering.
For inquiries, please use https://linkeddataorchestration.com/contact/
Meet the Host
I've got Tech, Data, AI and Media, and i'm not afraid to use them.
My name is George Anadiotis, and i am a writer, a planner and a doer. I am an Onalytica Top 100 Influencer in Big Data and Cloud, a Knowledge Graph expert, and a VentureBeat and ZDNet contributor, among other things.
Linked Data Orchestration is my brand. This podcast is where i share my work, as well as conversations with people who bring interesting news and views to the table.