Marco Rossetti, PhD

🚀 New Open Access Article 🚀 Emanuele Cavenaghi published his work on TORS! 👇 His paper, "Towards a Causal Decision-Making Framework for Recommender Systems", addresses the limitations of learning from offline observed data and the challenges of current debiasing strategies. To advance the field, he proposes integrating causality frameworks such as potential outcomes and structural causal models into recommender systems. This approach, developed in collaboration with Free University of Bozen-Bolzano, aims to enhance future research by applying causal discovery strategies on offline data, enabling the computation of counterfactuals and improving debiasing methods. You can find the full paper at: https://lnkd.in/dVgcJbMq Congratulations to Emanuele and his team for this amazing contribution! Dive into our research to explore these innovative ideas and their potential to transform recommender systems at: https://lnkd.in/dtpqzmX4 #Research #ArtificialIntelligence #MachineLearning #RecommenderSystems #Causality #MADLab

9

Read Marcello Politi's article to get a better understanding of Einstein’s notation and einsum multiplication 👉 #AI #MachineLearning #DataScience #DeepLearning

3

#OpenAI has just launched its latest #AI model, #OpenAI o1 (also known internally as Project Strawberry/Q*). What's different about OpenAI o1 is that it "thinks" before it answers—it can produce a long internal chain of thought before responding to the user. #OpenAI #AI #GenAI #Development #Research More details: https://lnkd.in/guMtKbg3 Simple test:

19

LinkedIn - the OG knowledge graph experts. They were doing KG integrations before it was cool! I love the emerging use case in seeing if using LLM’s as a scalable search quality rating vehicle, particularly something like SMOL’s could make financial sense to not just generate a 1:1 pairing of search to enriched grading but I could even see it making sense to have multiple enrichments per single query (1:N) based entirely on downstream activity (engagement flag for “long dwell” time as they would say).

3

2 commenti

#Meta has just unveiled Llama 3, their latest open-source large language model (LLM), and it's making waves across the AI landscape! 📊 Performance Insights: On popular benchmarks, it’s already beating out Claude 3 Sonnet, Gemini Pro 1.5, Mistral 7B Instruct & Gemma 7B-It. 🔧 Under the hood: -> 7x larger training data than Llama 2 -> 4x more code -> 30+ languages supported -> 8B & 70B parameter model versions 🌐 Availability: It's already available on the #AWS & #IBM WatsonX platform and will soon be accessible on these major platforms: ☁ #GCP, #Azure 🤗 #HuggingFace 🧠 #MetaAI assistant Read the full release here: https://lnkd.in/gdEKS9Xq #AI #GenAI #LLM #LLama-3 #OpenSource

11

2 commenti

Enlit was an exciting conference .. abuzz with Fast Data and Innovative Analytics solutions. I spoke about Visual Data Science and Event-Driven AI for optimization of energy production from wind turbines. The presentation featured the @Spotfire Gen-AI Copilot + Generating and Explaining Spotfire visuals + Generating Spotfire data science functions + Inentifying anomalies and getting to root cause We identified parts of the turbines that failed (eg controllers) in some regions, related that to certain suppliers; and showed significant cost savings, extension of equipment life and massive cost savings. I enjoyed learning about ETIP SNET — European Technology and Innocation Platform - Smart Networks for Energy Fransition I look forward to getting more involved in these innovative European Energy communities. @spotfire #VisualDataScience

27

Power of Agents and Agentic Flows in AI-Driven Systems As AI systems become more sophisticated, agents and agentic flows are emerging as the next evolution in automation and decision-making. These components are revolutionizing how businesses approach problem-solving, workflows, and dynamic environments. What Are Agents? AI agents are autonomous entities capable of perceiving their environment, making decisions, and taking actions to achieve specific goals. Agents continuously learn from interactions with their environment, allowing them to adapt to changing circumstances and optimize outcomes. Agentic Flows Explained: Agentic flows refer to the coordinated actions of multiple agents working together in a system to achieve complex objectives. These flows ensure that agents can communicate, collaborate, and align their efforts seamlessly across different tasks and processes. Agentic flows streamline how decisions are made and how tasks are executed, enabling more efficient and adaptive workflows. Key Benefits of Agents and Agentic Flows • Autonomy and Efficiency: Agents act independently, reducing the need for constant human intervention and enabling faster response times in dynamic environments. • Scalability: As businesses grow, agents and their flows can scale effortlessly, optimizing performance across vast systems and data landscapes. • Dynamic Decision-Making: Agents are capable of learning and adapting in real-time, making them ideal for decision-making in complex, fast-paced environments like telecoms, fraud management, and customer service. • Collaboration: Agentic flows allow agents to work in harmony, ensuring that distributed tasks are handled efficiently and holistically, with minimal resource waste. Real-World Applications: From automating complex business processes to handling distributed workflows in industries such as finance, telecoms, and logistics, agentic flows are paving the way for smarter, more resilient AI-driven systems. Whether it’s optimizing network traffic or fraud detection, agentic systems are transforming the way businesses operate by offering real-time, intelligent decision-making capabilities. The future is agent-driven, and leveraging agentic flows will be critical to building the next generation of intelligent, adaptive, and scalable solutions. #AI #Agents #AgenticFlows #Automation #ML #DynamicWorkflows #ScalableAI #AutonomousSystems #DecisionMaking #BusinessInnovation #AIinBusiness #FutureOfAI #agentsforfrauddetection #frauddetection #agentsforrevenueassurance #datascience #machinelearning #gpt #llama #mistral

20

📈 Dataframe libraries evolving faster than you can keep up? Marco Gorelli's Data Umbrella webinar is your cheat code! Learn Polars expressions and how to build library-agnostic data tools. 👀 Watch now: https://lnkd.in/dpMGfx4a #DataScience #TechTips

7

1 commento

"Strawberry" is here! 🍓 The new model from OpenAI can reason independently to solve more complex tasks. It's a new scaling paradigm that, instead of creating larger models, focuses on inference compute to question itself "behind the scenes" to produce better results.

11

The final post in my series on the Unified Star Schema (USS) by Bill Inmon and Francesco Puppini. This time I look at more benefits and come to a conclusion. I have looked at many different data modeling techniques during my career, and none have appealed as much as the USS. The benefits of adopting it are huge, especially for data teams looking to offer self-service access to data. I'm sure there will be edge cases where it doesn't suit, but I think it should be considered very broadly for adoption. It elegantly solves some of the biggest problems in data warehousing, and almost certainly reduces the total cost of ownership of Business Intelligence for any org. Thanks again to Technics Publications who have offered a discount code to buy a copy of the book, which is at the end of the post. https://lnkd.in/g9sumhDm

19

We are live with ツ Dr. Stefania Pecore, Independent AI and Data Science Expert. She is presenting, "Transforming Industries and Daily Life: The Power and Challenges of NLP and Generative AI"

12

As we approach 2025, generative AI is revolutionizing business operations by enhancing efficiency and customer experiences. 💡 Key trends include hyper-personalized marketing, where AI uses data insights for more targeted strategies, and intelligent automation, which streamlines processes like supply chain management and customer service to boost productivity. Conversational AI, with advances in natural language processing, is improving customer interactions, especially in support services. Multi-modal AI integrates diverse data forms for richer insights, benefiting various sectors. In healthcare, generative AI is transforming personalized treatment and drug discovery, while in cybersecurity, it enhances threat detection and data protection. Ethical AI development focuses on guidelines addressing bias, privacy, and transparency. Decentralized AI using blockchain strengthens data security, crucial for healthcare and finance. In creative fields, AI is innovating content creation, and in gaming and entertainment, it's providing immersive experiences. How do you see generative AI transforming your specific industry by 2025? Share in the comments! #GenerativeAI #AITrends #BusinessEvolution #DigitalInnovation

3

📣 Don't miss Marco Gorelli's talk at #PyConItaly on supporting multiple DataFrame libraries seamlessly.🙌 He'll cover the "just convert to pandas" limitations and how the DataFrame Interchange Protocol and Narwhals offer a better solution for robustness and efficiency. 🚀 https://2024.pycon.it/en #DataFrame #Polars #OpenSource

1

A Tale of Trust and Accuracy: Base vs. Instruct LLMs in RAG Systems Florin Cuconasu, Giovanni Trappolini, Nicola Tonellotto, Fabrizio Silvestri Abstract Retrieval Augmented Generation (RAG) represents a significant advancement in artificial intelligence combining a retrieval phase with a generative phase, with the latter typically being powered by large language models (LLMs). The current common practices in RAG involve using "instructed" LLMs, which are fine-tuned with supervised training to enhance their ability to follow instructions and are aligned with human preferences using state-of-the-art techniques. Contrary to popular belief, our study demonstrates that base models outperform their instructed counterparts in RAG tasks by 20% on average under our experimental settings. This finding challenges the prevailing assumptions about the superiority of instructed LLMs in RAG applications. Further investigations reveal a more nuanced situation, questioning fundamental aspects of RAG and suggesting the need for broader discussions on the topic; or, as Fromm would have it, "Seldom is a glance at the statistics enough to understand the meaning of the figures". 👉 https://lnkd.in/dxkpkqiS #machinelearning

1

AI is taking another leap forward—Anthropic, a rival to OpenAI, has just released a model that can do more than assist. It can use and control your computer. Imagine an AI that doesn’t just provide answers, but can see your screen, move your cursor, click, type, and explain its decisions. That's the latest from Anthropic, currently available in beta. This isn't just an interesting new functionality; it reinforces one of the most significant trends in AI development: agency. Agentic frameworks refer to AI models that can plan and execute actions step by step, using tools or interfaces to accomplish tasks—essentially combining reasoning with action. Over the past year, these frameworks have been gaining attention, and this is the first time I've seen it embedded natively in a model. What's particularly interesting is how AI assistants (chatbots) and enterprise process automation (RPA) are converging. Companies like UiPath are integrating generative AI into their automation suites, while AI labs are adding process automation to their capabilities—Anthropic's latest model being a prime example. The big question is: Which approach will dominate for enterprises and consumers? And in this evolving landscape, who will disrupt whom? https://lnkd.in/dhg-u9s2

49

4 commenti

#ThoughtLeadership 𝗔𝗱𝘃𝗮𝗻𝗰𝗲𝘀 𝗶𝗻 𝗦𝗲𝗺𝗶- 𝗦𝘂𝗽𝗲𝗿𝘃𝗶𝘀𝗲𝗱 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴 𝗮𝗻𝗱 𝗔𝗻𝗼𝗺𝗮𝗹𝘆 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 Recent research, led by ML Engineer Aniket S. at Unleash Networks, focuses on enhancing Trisul’s network monitoring capabilities by integrating deep-learning techniques for improved real-time traffic forecasting and top-k anomaly detection. 𝗘𝘅𝗽𝗹𝗼𝗿𝗲 𝗢𝘂𝗿 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀 𝗢𝗻 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗧𝗿𝗮𝗳𝗳𝗶𝗰 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴: ▪𝗫𝗚𝗕𝗼𝗼𝘀𝘁-𝗟𝗦𝗧𝗠 𝗳𝗼𝗿 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗧𝗿𝗮𝗳𝗳𝗶𝗰 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴: Discover how we combined two powerful models for accurate forecasting. ▪𝗧𝗼𝗽-𝗸 𝗔𝗻𝗼𝗺𝗮𝗹𝘆 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝗨𝘀𝗶𝗻𝗴 𝗖𝗼𝗻𝘃𝗔𝗩𝗔𝗘: Learn about our semi-supervised anomaly detection using custom variational autoencoders 🔗𝗥𝗲𝗮𝗱 𝗧𝗵𝗲 𝗙𝘂𝗹𝗹 𝗕𝗹𝗼𝗴 𝗛𝗲𝗿𝗲: https://lnkd.in/gZwMSSqa Stay tuned for more updates on latest advancements in 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗠𝗼𝗻𝗶𝘁𝗼𝗿𝗶𝗻𝗴 and 𝗧𝗵𝗿𝗲𝗮𝘁 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 with Trisul Network Analytics! #NetworkMonitoring #DeepLearning #SemiSupervisedLearning #Trisul #MachineLearning #AnomalyDetection #TrafficForecasting #XGBoost #LSTM #NetworkSecurity #ThreatDetection #NetworkAnalytics #ResearchAndDevelopment #SecurityAnalytics #mustread #BlogPost #FridayThoughts #FridayFocus #ResearchInsights

8

During searching for new ideas for first steps towards superintelligence I came to the intersection between cognitive evolution and LLMs, for example here the evolutionary algorithms (but given enough compute power one could conceive of simulating cognitive evolution and then increasing the simulation speed). I have not yet completely read the article, but it more and more goes in the direction, at least from the way of thinking, of OpenCog's Hyperon (https://lnkd.in/ek3HaNeX), which is of course way more sophisticated. #artificialIntelligence #cognitiveEvolution #superintelligence #openCog

2

#syntheticdata #lossofvariance #LLMs #SLMs “Initially, common types such as golden retrievers would dominate while less common breeds such as Dalmatians disappeared. Finally, the images of golden retrievers themselves would become an anatomic mess, with body parts in the wrong place.” https://lnkd.in/en3mrajb

1