VERSES Challenges AI Industry with Benchmark Tests
- VERSES AI Inc. aims to demonstrate superior performance in AI models using less data and energy.
- The company's research roadmap highlights milestones in image classification, Atari gaming challenges, and multi-agent systems.
- VERSES plans to release results showcasing efficiency and innovation in Q1–Q2 2024 for the first benchmark.
- The second benchmark, Atari 10k Challenge, aims to demonstrate vast sample and compute efficiency compared to alternatives.
- VERSES' active inference-based AI shows promise in achieving human-level performance with minimal data usage.
- The third benchmark, NeurIPS 2024 Melting Pot Challenge, focuses on multi-agent systems and uncertain environments.
- VERSES intends to showcase the benefits of active inference agents in engineering smart multi-agent systems.
- None.
VANCOUVER, British Columbia, Feb. 22, 2024 (GLOBE NEWSWIRE) -- VERSES AI Inc. (CBOE:VERS) (OTCQB:VRSSF) (“VERSES” or the “Company”), a cognitive computing company developing next-generation intelligent software systems, today provides a research roadmap that outlines the key milestones and benchmarks against which to measure the progress and significance of the Company’s research and development efforts, against conventional deep learning, for the benefit of industry, academia, and the public.
“We laid out a roadmap that can be accessed at https://www.verses.ai/rd-overview, which we expect to use to demonstrate over the course of this year that VERSES’ approach to AI is able to match or exceed the performance of advanced AI models on multiple industry-standard benchmarks while using materially less data and energy," said Gabriel René, founder and CEO of VERSES.
This is notable in light of OpenAI’s CEO Sam Altman’s recent statement that the future of AI depends on an energy breakthrough1 along with a plan to raise
Mr. René further stated, “The implications of meeting these benchmarks is to provide scientific evidence that VERSES’ approach can yield better, cheaper and faster AI that applies to a broader market opportunity and is commercialized in our Genius Platform. We have published our research roadmap so that both the industry and the public can track our progress.”
____________________________
1 https://www.reuters.com/technology/openai-ceo-altman-says-davos-future-ai-depends-energy-breakthrough-2024-01-16/
2 https://www.wsj.com/tech/ai/sam-altmans-vision-to-remake-the-chip-industry-needs-more-than-money-1dc0678a
First benchmark: Classification and generation tasks
With the first benchmark, VERSES intends to demonstrate the compute and sample efficiency on image classification and generation tasks such as MNIST and CIFAR; in particular, demonstrating the computational efficiency of VERSES’ approach over and above other modern Bayesian inference toolboxes, such as NumPyro. We also intend to show how this approach is competitive with the computational efficiency of traditional deep learning approaches based on tools like PyTorch—but augmented with the great sample efficiency that comes from adopting a fully Bayesian approach. The Company plans to release these results demonstrating the efficient compute and improved sample efficiency of our approach to classification and generation tasks around the end of Q1–Q2 2024 in open-access publications.
Second benchmark: Atari 10k Challenge
With the second benchmark, the Atari 10K Challenge, VERSES intends to demonstrate that its approach is vastly more sample and compute efficient than other alternatives. The initial Atari benchmark challenge was introduced in 2015 and involved producing a single AI system that could meet or beat human-level performance on 26 classic Atari games. The AI model must learn directly from pixel data, using only the score as a reward signal. The initial architecture designed for this was data-heavy, using years of gameplay—usually more data than a human player might ever have access to.
To address this, the Atari 100k benchmark was introduced, which restricts the amount of gameplay used in learning to 100,000 environment steps. Atari 100k is a good benchmark to showcase the power and sample efficiency properties of the active inference approach. The Company expects to demonstrate two sources of gains in efficiency. The first comes from fast online learning of the world model for the game. The second comes from efficient policy estimation that does not require periodic resets of the sort used by traditional gradient-based methods, such as Q-learning.
Although the Atari 100k (2 hours of gameplay) is the industry-leading benchmark, and VERSES plans to demonstrate competitive play at the 100k benchmark, the Company intends to further showcase the unique strengths of active inference-based AI, namely, rapid learning and improved sample efficiency by proposing the Atari 10k benchmark challenge (roughly 12 minutes of gameplay), using only raw pixel data and the score as input. The challenge is to reach human-level performance (or greater) measured on the same amount of gameplay. Humans can achieve competent play very quickly, but how do advanced architectures perform? VERSES intends to demonstrate that our system can outperform sophisticated deep learning on the 10k benchmark—learning to play the game efficiently with little data. Our preliminary results currently demonstrate that our agents are able to learn the dynamics of gameplay and score on simple games in only several thousand steps, demonstrating more efficient learning using a model that is ninety-nine percent smaller in parameter size than the leading competitors, and able to train on a laptop without a large GPU infrastructure.
The Company plans to share final results in Q3 2024, as well as in open-access publications.
Third benchmark: NeurIPS 2024 Melting Pot Challenge
The previous two benchmarks cater to the strengths of deep learning approaches, i.e., they often involve noiseless tasks that are completely observed (with no ambiguity) and that involve well-defined reward functions.
These benchmarks do not showcase the power of active inference. For the third benchmark, VERSES intends to use the new multi-agent NeurIPS Melting Pot Challenge benchmark since the ultimate goal is to develop more naturalistic benchmarks that showcase the ability of active inference agents to deal with uncertain environments. Specifically, one of the main advantages of building active inference agents that work directly in belief space with an explicit representational structure is that it becomes possible to share beliefs between agents.
The Company believes that this benchmark will showcase the benefits that active inference brings for engineering multi-agent systems and align with the central ambitions of VERSES AI research: to create ecosystems of AI systems.
VERSES plans to share these results showcasing the unique ability of active inference agents to lay the foundations of smart multiagent systems around Q4 2024–Q1 2025, additionally in open-access publications.
About VERSES
VERSES AI is a cognitive computing company specializing in biologically inspired distributed intelligence. Our flagship offering, Genius™, is patterned after natural systems and neuroscience. Genius™ can learn, adapt and interact with the world. Key features of Genius™ include generalizability, predictive queries, real-time adaptation and an automated computing network. Built on open standards, Genius™ transforms disparate data into knowledge models that foster trustworthy collaboration between humans, machines and AI, across digital and physical domains. Imagine a smarter world that elevates human potential through innovations inspired by nature. Learn more at VERSES, LinkedIn and X.
On behalf of the Company
Gabriel René, Founder & CEO, VERSES AI Inc.
Press Inquires: press@verses.ai
Investor Relations Inquiries
U.S., Matthew Selinger, Partner, Integrous Communications, mselinger@integcom.us 415-572-8152
Canada, Leo Karabelas, President, Focus Communications, info@fcir.ca 416-543-3120
Forward Looking Information
This press release contains "forward-looking information" and "forward-looking statements" within the meaning of applicable securities legislation (collectively, “forward-looking statements”). The forward-looking statements herein are made as of the date of this press release only, and the Company does not assume any obligation to update or revise them to reflect new information, estimates or opinions, future events or results or otherwise, except as required by applicable law. Often, but not always, forward-looking statements can be identified by the use of words such as "plans", "expects", "is expected", "budgets", "scheduled", "estimates", "forecasts", "predicts", "projects", "intends", "targets", "aims", "anticipates" or "believes" or variations (including negative variations) of such words and phrases or may be identified by statements to the effect that certain actions "may", "could", "should", "would", "might" or "will" be taken, occur or be achieved. These forward-looking statements include, among other things, statements relating to: the expectation that Verses will use the roadmap to demonstrate over the course of this year that VERSES’ approach to AI is able to match or exceed the performance of advanced AI models on multiple industry-standard benchmarks while using materially less data and energy; that VERSES intends to demonstrate its compute and sample efficiency on image classification and generation tasks such as MNIST and CIFAR; that Verses intends to show how this approach is competitive with the computational efficiency of traditional deep learning approaches based on tools like PyTorch; that Verses plans to release the first benchmark’s results around the end of Q1–Q2 2024 in open-access publications; that Verses expects to demonstrate with the second benchmark that VERSES’ approach is vastly more sample and compute efficient than other alternatives through two sources of gains in efficiency; that VERSES plans to demonstrate competitive play at the 100k benchmark; that Verses intends to showcase the unique strengths of active inference-based AI, namely, rapid learning and improved sample efficiency using little data through the Atari 10k benchmark challenge; that Verses plans to share final results of the second benchmark in Q3 2024 in open-access publications; that VERSES intends to use a third benchmark based on the new multi-agent NeurIPS Melting Pot Challenge to showcase the ability of active inference agents to deal with uncertain environments; that VERSES plans to share the results of the third benchmark around Q4 2024–Q1 2025 in open-access publications.
Such forward-looking statements are based on a number of assumptions of management, including, without limitation: that Verses will successfully use the roadmap to demonstrate over the course of this year that VERSES’ approach to AI is able to match or exceed the performance of advanced AI models on multiple industry-standard benchmarks while using materially less data and energy; that VERSES will demonstrate its compute and sample efficiency on image classification and generation tasks such as MNIST and CIFAR; that Verses will show how this approach is competitive with the computational efficiency of traditional deep learning approaches based on tools like PyTorch; that Verses will release the first benchmark’s results around the end of Q1–Q2 2024 in open-access publications; that Verses will demonstrate with the second benchmark that VERSES’ approach is vastly more sample and compute efficient than other alternatives through two sources of gains in efficiency; that VERSES will demonstrate competitive play at the 100k benchmark; that Verses will showcase the unique strengths of active inference-based AI, namely, rapid learning and improved sample efficiency using little data through the Atari 10k benchmark challenge; that Verses will share final results of the second benchmark in Q3 2024 in open-access publications; that VERSES will use a third benchmark based on the new multi-agent NeurIPS Melting Pot Challenge to showcase the ability of active inference agents to deal with uncertain environments; that VERSES will share the results of the third benchmark around Q4 2024–Q1 2025 in open-access publications.
Additionally, forward-looking statements involve a variety of known and unknown risks, uncertainties and other factors which may cause the actual plans, intentions, activities, results, performance or achievements of the Company to be materially different from any future plans, intentions, activities, results, performance or achievements expressed or implied by such forward-looking statements. Such risks include, without limitation: that Verses will not use the roadmap to demonstrate over the course of this year or at all that VERSES’ approach to AI is able to match or exceed the performance of advanced AI models on multiple industry-standard benchmarks or any benchmarks while using materially less data and energy; that VERSES will not successfully demonstrate its compute and sample efficiency on image classification and generation tasks such as MNIST and CIFAR; that Verses will not successfully show how this approach is competitive with the computational efficiency of traditional deep learning approaches based on tools like PyTorch; that Verses will not release the first benchmark’s results around the end of Q1–Q2 2024 in open-access publications or at all; that Verses will not successfully demonstrate with the second benchmark that VERSES’ approach is vastly more sample and compute efficient than other alternatives through two sources of gains in efficiency or any at all; that VERSES will not demonstrate competitive play at the 100k benchmark; that Verses will not showcase the unique strengths of active inference-based AI, namely, rapid learning and improved sample efficiency using little data through the Atari 10k benchmark challenge; that Verses will not share final results of the second benchmark in Q3 2024 in open-access publications or at all; that VERSES will not successfully use a third benchmark based on the new multi-agent NeurIPS Melting Pot Challenge to showcase the ability of active inference agents to deal with uncertain environments; that VERSES will not share the results of the third benchmark around Q4 2024–Q1 2025 in open-access publications or at all.
The forward-looking statements contained in this press release represent management's best judgment based on information currently available. No forward-looking statement can be guaranteed and actual future results may vary materially. Accordingly, readers are advised not to place undue reliance on forward-looking statements. Neither the Company nor any of its representatives make any representation or warranty, express or implied, as to the accuracy, sufficiency or completeness of the information in this press release. Neither the Company nor any of its representatives shall have any liability whatsoever, under contract, tort, trust or otherwise, to you or any person resulting from the use of the information in this press release by you or any of your representatives or for omissions from the information in this press release.
FAQ
What is VERSES AI Inc. aiming to demonstrate with its research roadmap?
When does VERSES plan to release results for the first benchmark?
What is the focus of the second benchmark, Atari 10k Challenge?
What is the goal of the third benchmark, NeurIPS 2024 Melting Pot Challenge?
What unique strength does VERSES' active inference-based AI demonstrate?
