Author: Kannan Keeranam, Director, Cloud & AI Strategy and Execution at Intel
intel announced it has strengthened its strategic research and co-innovation collaboration with Mila, a Montreal-based artificial intelligence research institute. As part of a renewed three-year commitment, Intel and Mila researchers will focus on responsible AI and the development of advanced AI techniques for problems such as climate change, new materials discovery and digital biology. Solving these complex challenges requires extensive AI research and a commitment to open innovation to drive state-of-the-art (SOTA) AI.
Automated AI-driven discovery of novel materials
As scientists and organizations work to fight climate change, one of the most relevant research topics is the AI-driven discovery of novel materials that can drastically reduce the cost and carbon footprint of technologies. While advances in chemical simulation techniques, such as density functional theory, have developed various methods capable of simulating important properties of complex material systems, these techniques have been limited in the complexity of the material systems they can model due to the unfavorable scaling of computational costs with increasing number of atoms. AI techniques, particularly graphical neural networks (GNNs), have made advances attributed to their ability to approximate chemical simulations with significantly reduced computational effort, especially as system size increases. This holds great promise for using AI-assisted simulated techniques to replicate material systems with greater complexity and applicability to modern technological and societal challenges such as climate change.
As part of this commitment, Intel and Mila will work together to develop scientific and technological innovations to improve the performance of GNNs in atomistic simulations like this Open the Catalyst record. These efforts can democratize the ability of researchers to engage with atomistic materials data by improving the associated technology pipeline. Research teams will work to create learning-based frameworks to enhance search in material design applications. These frameworks can draw on ideas from reinforcement learning, search algorithms, generative models, as well as other machine learning algorithms, including generative flow networks developed by Mila researchers. The main focus of this research path is the development of algorithms for materials design challenges as well as the development of ecosystems by creating toolkits for common challenges faced by researchers.
In addition, teams from Intel and Mila will apply NLP (Natural Language Processing) techniques to journal articles, articles, websites and/or patents. This will allow researchers to collect and apply technical knowledge contained in texts to discover new systems of materials, understand the various physical and chemical phenomena involved in materials synthesis to accelerate discovery pathways, and develop methods to model structured Create data from unstructured data. The knowledge gained from this line of research will provide valuable synergies for the above research efforts to significantly advance AI-driven material design.
Causal machine learning for climate science
Standard physics-based climate models can help predict the effects of climate change, but they are complex and computationally intensive. They often take months to execute—even on specialized supercomputing hardware—reducing the frequency of simulation runs and the ability to develop granular, localized predictions.
Intel believes that global climate change is a serious environmental, economic, and social challenge that requires an equally serious response from governments and the private sector. The company has for years taken active steps to reduce its own environmental footprint and recently announced plans to achieve net-zero greenhouse gas emissions from its global operations by 2040. With today’s announcement, Intel and Mila join forces to further address this challenge through the development and application of new and advanced AI techniques. With this collaboration, Intel and Mila will build a new breed of climate model emulator based on causal machine learning (ML) to identify which variables are predictive from extremely high-dimensional inputs to traditional climate models. The teams will start with probabilistic predictions of regional precipitation and temperatures, learning from ensembles of climate model simulations CMIP6 data archive. The next step will be to develop more sophisticated algorithms to work with regional climate drivers such as land cover changes.
Ultimately, the project will enable significant advances in climate science and inform policy directly by enabling local and regional predictions of climate change impacts. The project will also advance causal ML due to the large number of relevant variables and complex interrelationships between them with a relatively large number of causal links and numerous confounders.
Digital Biology: Accelerating Research into the Molecular Causes of Disease and Drug Discovery
Biology is an exciting frontier in the natural sciences. With the availability of high-resolution data, the emergence of breakthroughs in AI, and the growth in computing density driven by Moore’s Law, it’s coming to the computing realm now more than ever. The time has come to usher in the era of precision medicine, where we learn from everyone’s data to benefit everyone.
To achieve this vision, Intel and Mila will develop AI techniques to:
- Understanding the molecular drivers behind diseases. Predicting complex phenotypes, including diseases, based on the genotype of single nucleotide polymorphisms (SNPs), formerly known as Polygenic Risk Score (PRS) prediction, has long been a challenge in digital biology. Since most phenotypes are influenced by many SNPs throughout the genome, the main computational challenge is to collectively learn the causal effects of all SNPs in the genome on the phenotypes, using large population data. With millions of SNPs discovered so far (e.g. UK Biobank data set), an exact solution is computationally unsolvable.
- Identify the most promising drug molecules. It lasts more than 10 years and $2.5 billion to develop a new drug and Identify the most promising target drug molecules. This offers a great opportunity for ML-based methods. With the amount of data already being generated and graphed in the biomedical field (e.g. scientific literature, known drug molecules, drug-protein bonds, etc.), there is an opportunity to develop open-source ML frameworks for the use drug research. For example, TorchDrug, https://torchdrug.ai/Mila’s ML framework for drug discovery based on Graph ML, Deep Generative Models and Reinforcement Learning can be used to significantly reduce the cost and time to market of drug development.
These and many similar problems of broad scientific interest are computationally difficult to tackle when we try to find exact solutions, but can be characterized as learning problems: discovery of gene regulatory networks as causal discovery and drug discovery as active learning. With this approach, Intel and Mila will take on these challenges together to build novel high-performance methods for 1) causal AI methods and 2) AI algorithms to identify promising drug molecules and usher in the long-promised era of precision medicine.
Notices and Disclaimers
Your costs and results may vary.
Intel does not control or verify third-party data. You should consult other sources to assess accuracy.
Statements in this document that relate to future plans or expectations are forward-looking statements. These statements are based on current expectations and involve many risks and uncertainties that could cause actual results to differ materially from those expressed or implied by such statements. For more information on the factors that could cause actual results to differ materially, please see our most recent earnings announcement and SEC filings at www.intc.com.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and marks may be claimed as the property of others.