Bandicoot_Inv
2 days ago
VERSES Genius™ Agent Outperforms Leading AI Algorithms at Major Industry Benchmark
Demonstrating High Performance in “Atari Challenge” with 90% Less Data, 96% Faster Training, 96% Smaller Model Size; Applicable to Real-World Challenges
https://www.globenewswire.com/news-release/2025/01/22/3013198/0/en/VERSES-Genius-Agent-Outperforms-Leading-AI-Algorithms-at-Major-Industry-Benchmark.html
VANCOUVER, British Columbia, Jan. 22, 2025 (GLOBE NEWSWIRE) -- VERSES AI Inc. (CBOE:VERS) (OTCQB:VRSSF) ("VERSES'' or the "Company”), a cognitive computing company specializing in next-generation intelligent systems, has shared preliminary results on its variant of the “Atari 100K Challenge” a leading AI industry benchmark, where software agents must learn gameplay proficiency, on their own, using a limited amount of training data.
The results, which can be found in a blog post on the Company’s website titled “Mastering Atari Games with Natural Intelligence,” showcase how agents powered by GeniusTM, a toolkit for developing intelligent agents, were able to match or exceed the performance of state-of-the-art Deep Reinforcement Learning (DRL) and Transformer algorithms using 90% less data. DRL is the algorithm that underpins work such as Google Deepmind’s AlphaZero, AlphaGo, and AlphaFold, while Transformers are the foundation of Generative AI and LLMs like OpenAI’s GPT, and other popular models like Anthropic’s Claude, Google’s Gemini, Meta’s Llama, X.AI’s Grok and others.
“The Atari Challenge represents more than just machines playing games. Video games are a proxy for the complex dynamic systems all around us,” said Gabriel René, founder and CEO of VERSES. “We need AI to help us to better navigate the complexity and uncertainty of the real-world; yet state-of-the-art AI algorithms remain too unreliable, inefficient and unexplainable. While there is more work to be done to productize this research into Genius, we believe these early results and the research behind them signal a historical shift towards developing smarter, safer and more scalable AI.”
The Atari 100k benchmark has been used to test an agent's ability to excel in three critical areas: interactivity, generalization, and efficiency. VERSES’ variant, known as “Atari 10k,” tests for these same capabilities, using only 1/10th of the sample data. This increase in efficiency reduces the need to rely on large datasets and compute architectures, and indicates applicability to real-world problems where data can be sparse, incomplete, noisy, and where learning may need to occur in real time. Genius agents have demonstrated these capabilities across multiple Atari games by efficiently learning about the objects and physical mechanics of the game environments through interaction, on their own, without the need for human supervision. By learning the hidden cause-effect dynamics in the games, the agents were able to better predict outcomes and select optimal actions to outperform the state-of-the-art models even across multiple games with different objects, dynamics and goals. In contrast, DRL and Transformer architectures must rely on substantial training data and compute as well as bespoke tuning to ‘fit’ with the gameplay mechanics.
“By aligning with how intelligence has evolved in nature and how energy efficient biological systems must act to preserve their existence, we believe that the architecture and methods that underwrite Genius offer a sustainable and scalable means to building ecosystems of authentic, autonomous and agentic intelligence,“ said Karl Friston, VERSES Chief Scientist.
“By Applying the Free Energy Principle, the Active Inference framework, and Bayesian Machine Learning, we are working on developing an efficient and generalized architecture that allows intelligent agents to learn and develop expertise across any domain,“ said Hari Thiruvengada, VERSES Chief Technology Officer. “The reasoning, planning, and online learning capabilities that allow a Genius Agent to engage in dynamic gameplay is extremely relevant to applications such as classification, recommendations, prediction, and decision-making across a variety of industries, including financial, medical, risk analysis, autonomous driving, robotics and more.”
For a more detailed description and videos of gameplay visit our blog at verses.ai.
Bandicoot_Inv
2 weeks ago
VANCOUVER, British Columbia , Jan. 13, 2025 (GLOBE NEWSWIRE) -- VERSES AI Inc a cognitive computing company specializing in next generation intelligent systems, today announces that Chief Scientist Karl Friston and members of their senior leadership team will be returning to The World Economic Forum Annual Meeting 2025, taking place on January 20-24th in Davos, Switzerland , where they will participate in a number of meetings, panels and interviews.
During this event VERSES plans to release an update on the previously announced Atari 10K Challenge. Preliminary results have demonstrated that VERSES Active Inference-based models can achieve comparable or better performance than current state-of-the-art reinforcement learning models using just 90% less data (10k frames) and a fraction of the compute. The update will contain results and video demonstrations of VERSES meeting or exceeding human-level performance on multiple Atari games.
robinandthe7hoods
3 weeks ago
If you do a lot of reading,even industry experts admit they have no idea what they'll be able to do with LLMs and how to make them profitable.
Even at this stage,they are an interesting toy.If they raise the price to play around with them,who would be willing to pay for it?
Businesses are still tinkering with them,knowing they aren't trustworthy enough to be left on their own,for the most part.
But everybody wants to build even more AI data centres for LLMs.Even Keven O'Leary is proposing a $70 billion centre in Alberta.
Current AI is a 2D configuration,meanwhile everyone knows the next generation must go to 3D and 4D.
It is also admitted by industry insiders that current AI can't reach AGI as is.
But,along comes Verses AI,which is 3D and 4D based AI,and which operates at a much lower cost base than LLMs,AND CAN ACTUALLY THINK!
It appears that true AGI is within reach now,by applying the power of the LLMs with the reasoning and cost savings of Verses.
It may also appear like the LLM operators are ignoring Verses at the moment.
Imo,by the end of this year,they will all be coming on board to add Verses 'Genius' to their own models,so they can finally finish building their product ahead of time, but also have it be profitable earlier,while entering the 3D and 4D format.
Genius is adaptable over different product lines and formats,while LLMs on their own are not.
This year will be interesting,while we wait for them to come to us,which WILL happen.
Either the LLMs partner with Verses,or Verses has to go into building their own computer/phone lines and LLM data centers to get the entire potential out of this product
These are the only two options.
Partnerships/buyouts with/for LLMs are coming.
These are the only two options for both sides.
Which LLMs are going to survive,and which are going to die?
robinandthe7hoods
3 weeks ago
Looking at over $2.00 today.If there was a $5-10-15 million PP coming the stock would have reversed toward the target price by now after this run.
Anybody thinking institutions can't get involved below $5.00 might be forgetting about those that deal penny stocks all the time.
"You help us now,and we'll let in on the big round in the overbought side"
If this run continues,then it's clear they have a group running this up,knowing it's all profit heading towards a NASDAQ listing and a minimum $4.00 price and a big funding round to prop it up coming.
If the goal here is NASDAQ,then they have to run it up to $8-10 to be safe from a selloff when it first starts to trade.
We'll see if this continues or a normal pullback takes place.
The short number is rising,so they're getting this will break,but a lot of news is set to be released,which might break the shorts.
One way or the other,the next week will be interesting to watch this.
You keep expecting this to calm down and reverse some,but it just keeps running.
Nothing says it's not a Blue Chip acquiring through multiple houses,no one knows... It's Germany,US and Canada all at the same time.
Bandicoot_Inv
3 weeks ago
VERSES Provides Q4 Corporate Update
VANCOUVER, British Columbia, Dec. 31, 2024 - VERSES AI Inc. ("VERSES'' or the "Company”), a cognitive computing company specializing in next generation intelligent systems, provides a corporate update.
“Our mission is to unleash a new class of intelligent autonomous agents that are both more reliable and more efficient. Today, the research and product investments we’ve made are proving their worth—just as the diminishing returns of scaling data and compute, alongside the unreliability of traditional deep and reinforcement learning, are becoming widely recognized,” said Gabriel René, founder and CEO of VERSES.
“Our Genius product has evolved rapidly this quarter, and we are excited to transition from beta to its commercial phase in 2025,” René continued. “Its unique combination of active inference and learning, as highlighted in today’s announcement on the Atari benchmarks, enables a new class of agents capable of reliably automating real-time decision-making, reducing errors, and maximizing efficiency for enterprises—from software applications to robotic automation. By design, Genius goes beyond intelligent systems available today and has the potential to become a disruptive force in next-generation autonomous intelligent systems,” concluded Mr. René.
Quarterly Highlights include:
Genius™ Rollout
"Being able to explicitly model the cause-effect relationships of complex systems and quantify uncertainty means we can generate something not possible with traditional ML tools – results that are reliable, explainable, and assurance ready," said Andy Tasker, CEO of Prodigii.
VERSES released an update of its flagship product, Genius, to beta program participants earlier this month, which includes powerful, enhanced reasoning and planning capabilities, comprehensive model explainability, and support tools to accelerate onboarding. The latest release improves user experience by providing a new model editing interface with simplified tutorials and deployment processes based on initial feedback from early beta users. This release is initially available to select beta partners and machine learning practitioners. As Genius functionality evolves we expect to expand the number and variety of user personas and use cases in 2025.
Research Milestones
“As anticipated, we’re thrilled to share groundbreaking early results from the Atari 10k Challenge,” said Hari Thiruvengada, Chief Technology Officer of VERSES.
VERSES Atari 10K Challenge is an aggressive variant of the Atari 100K benchmark that aims to meet or exceed human-level performance across multiple Atari games using just 10k frames (90% less data) and a fraction of the compute with a single generalizable architecture than industry-leading approaches. Preliminary results demonstrate that VERSES Active Inference-based, Bayesian models can achieve comparable or better performance than state-of-the-art reinforcement learning models such as the #1 ranked BBF (Bigger, Better, Faster) from Google’s Deepmind and other deep learning transformer-based models that power OpenAI, Anthropic and Meta’s to LLM’s and all Generative AI models across multiple games using 90% less data, 96% less compute and resulting in models that are up to 97% smaller in size, which is key to operating on low-power devices. In addition, VERSES model performance was conducted on the equivalent of a CPU vs the GPU architecture required for the other algorithms.
“We believe this marks an historic shift for the AI industry at large,” stated Thiruvengada. “We have demonstrated that it is technically feasible to leverage Bayesian approaches at scale, tackling complex challenges like Atari with significantly reduced compute and sample sizes. Simultaneously, we are integrating these advanced capabilities into Genius, ensuring that we deliver a user-friendly and accessible solution for our customers. This is just the beginning of what we can accomplish together.”
A more comprehensive update on Atari results is forthcoming.
The Atari benchmarks come on the heels of VERSES sharing results demonstrating how an agent, powered by Genius, outperformed OpenAI’s o1-Preview, considered the industry-leading reasoning model on the code-breaking test Mastermind. Genius consistently outperformed o1-Preview by reliably solving the code 100% of the time vs. o1-Preview’s 71% accuracy rate while performing 140 times faster and costing 5260 times less. A detailed breakdown of the results can be found here.
Commercial Projects
The initial smart city project in partnership with Analog for simulating taxi fleet management in Abu Dhabi is progressing well and the discovery phase is on schedule.
“We are pleased that a majority of Genius beta users are expected to convert to commercial engagements in 2025,” said James Hendrickson, Chief Operating Officer. “We also have several new projects in a variety of different market segments that are in the scoping phase with contracting expected in Q1.”
Corporate and Operational Updates
As the Company anticipates scaling product and commercial operations, we have made several investments to shore up personnel and organizational processes.
VERSES extended Professor Karl Friston’s exclusive contract as Chief Scientist. The multi-year contract builds on the success of both research and product development and includes incentives tied to future milestones and deliverables.
VERSES researchers attended the 38th Annual Conference on Neural Information Processing Systems (NeurIPS 2024) held in Vancouver, Canada. Three full papers from VERSES were accepted and seven were included in various workshops with topics spanning, among other areas, active inference, predictive coding, embodied agents, and robotics. Professor Karl Friston presented at the NeuroAI workshop entitled “The three faces of AI,” and participated in a panel discussion on the fusion of AI and Neuroscience with fellow AI pioneer Yoshua Bengio.
Additionally, the Company operationalized research with standardized test environments, frameworks and more robust support for cloud services like Amazon Web Services. There has also been a focus on formalizing and improving knowledge transfer efficiency between research and product in order to help prioritize research efforts based on demand as well as the productization of research.
https://www.globenewswire.com/news-release/2024/12/31/3003120/0/en/VERSES-Provides-Q4-Corporate-Update.html