Venture capital was pouring into new generative artificial intelligence technology before President Biden issued an order that could cool the rate of funds from flowing into the sector. Experts say that's not necessarily a bad thing for bank use of AI.

Should banks fear — or favor — a slowdown in AI investment?

Embedded financial services have proven ... entering the marketplace for the first time. “Software providers, in particular, are offering embedded finance services by leveraging external customer data, next-generation decisioning models and API-enabled ...

Embedded lending: helping consumers through a rough economy

If adopted correctly, financial services ... Officer at First Abu Dhabi Bank, continues: “Outside of risk management, quantum supercomputing will lead to a range of banking capabilities, such as analysing large areas of unstructured data to make ...

How quantum computing could transform the banking sector

First Business Financial Services (FBIZ) declares $0.2275/share quarterly dividend, in line with previous. Forward yield 3.07% Payable Nov. 16; for shareholders

First Business Financial Services declares $0.2275 dividend

Operator: Good morning. My name is Rob, and I will be your conference operator today. At this time, I would like to welcome everyone to the Provident Financial Services, Inc.

Provident Financial Services, Inc. (NYSE:PFS) Q3 2023 Earnings Call Transcript

For the quarter ended September 2023, First Business Financial Services (FBIZ) reported revenue of $37.03 million, up 8.6% over the same period last year. EPS came in at $1.17, compared to $1.25 in the year-ago quarter.

Here's What Key Metrics Tell Us About First Business Financial Services (FBIZ) Q3 Earnings

MADISON, Wis. (AP) — MADISON, Wis. (AP) — First Business Financial Services Inc. (FBIZ) on Thursday reported third-quarter earnings of $9.9 million. The Madison, Wisconsin-based bank said it had earnings of $1.17 per share. The results missed Wall ...

First Business Financial Services: Q3 Earnings Snapshot

C&F Financial Corporation (the Corporation) (NASDAQ: CFFI), the one-bank holding company for C&F Bank, today reported consolidated net income of $5.8 million for the third quarter of 2023, which represents a decrease of $768,

C&F Financial Corporation Announces Net Income for Third Quarter and First Nine Months

As per Verizon’s 2023 Data Breach Investigation Report, the financial services industry was the most targeted industry, with compromised personal data accounting for almost 74% of the industry ...

Decoding the importance of data protection for financial institutions

The financial industry is increasingly adopting AI to offer personalised services. AI algorithms analyse individual customer data to tailor financial products and services to specific needs.

AI in finance: aiding data-driven financial decision-making

Get the most recent info and news about First Data on HackerNoon, where 10k+ technologists publish stories for 4M+ monthly readers.

Fiserv, Inc., together with its subsidiaries, provides payment and financial services technology worldwide. The company operates through Acceptance, Fintech, and Payments segments. The Acceptance segment provides point-of-sale merchant acquiring and digital commerce services; mobile payment services; security and fraud protection products; Carat, an omnichannel commerce solution; Clover, a cloud-based point-of-sale and business management platform; and Clover Connect, an independent software vendors platform. This segment distributes through various channels, including direct sales teams, strategic partnerships with agent sales forces, independent software vendors, financial institutions, and other strategic partners. The Fintech segment offers customer deposit and loan accounts, as well as manages an institution's general ledger and central information files. This segment also provides digital banking, financial and risk management, professional services and consulting, item processing and source capture, and other products and services. The Payments segment offers card transactions, such as debit, credit, and prepaid card processing and services; security and fraud protection products; card production; print services; and various network services, as well as non-card digital payment software and services, including bill payment, account-to-account transfers, person-to-person payments, electronic billing, and security and fraud protection products. It serves business, banks, credit unions, other financial institutions, merchants, and corporate clients. The company was incorporated in 1984 and is headquartered in Milwaukee, Wisconsin.

First Data

When we think about data networks and systems that use mechanical mechanisms to send data across a distance, we think about this concept as a pretty modern one that only existed as a practical idea in the late 1900s. However, what if I told you that one of the earliest data networks to exist predated the modern internet as we know it? Heck, this ancient data network predates modern digital computers and analog computer systems. The modern electric telegraph is still younger than this particular early data network and most mechanical computers have not been around as long as this old-school data network. And when I say old-school, I mean truly old-school, as this data network was around before the Medieval Times and was developed by one of the oldest civilizations in written history. There is an ancient device known as the telegraph. This particular device was created to use water and devices in order to send information over long distances. The earliest version of this creation was produced back around 350 BCE by an early inventor and ancient Greek writer by the name of Aeneas Tacticus. It was produced by Tacticus as a device to be used for some long-form data communications during wartime & includes a very complicated system. The telegram worked by having identical containers on separate hills that are not connected to one another with each container being filled with water & having a vertical rod floating within both containers. The rods would have a bunch of predetermined codes put on them at various points along these poles. There would be a spigot at the bottom of these containers and water could be drained out of the device until a desired code was reached. Using a torch or signal, the person sending the code would be able to indicate how long it takes for the water to drain out to the desired code, allowing the receiver to know what signal/code is being communicated. hydraulic mechanical How these particular telegraph devices work is even explained by the Greek historian Polybius in his book : The Histories Then we are to fill the vessels with water and put on the corks with the rods in them and allow the water to flow through the two apertures. When this is done it is evident that, the conditions being precisely similar, in proportion as the water escapes the two corks will sink and the rods will disappear into the vessels. When by experiment it is seen that the rapidity of escape is in both cases the same, the vessels are to be conveyed to the places in which both parties are to look after the signals and deposited there. Now whenever any of the contingencies written on the rods occurs he tells us to raise a torch and to wait until the corresponding party raises another. When both the torches are clearly visible the signaler is to lower his torch and at once allow the water to escape through the aperture. Whenever, as the corks sink, the contingency you wish to communicate reaches the mouth of the vessel he tells the signaler to raise his torch and the receivers of the signal are to stop the aperture at once and to note which of the messages written on the rods is at the mouth of the vessel. This will be the message delivered, if the apparatus works at the same pace in both cases. What is so important about these primitive methods of sending data? Well, while it was limited in codes, it was a great early attempt at long-range telegraphic communication. It came as an early telegraph system that even comes before mechanical optical telegraphs and even inspired a British version that, while never made commercially available, was in 1838 by British civil engineer Francis Whishaw using pipes and more modern mechanical pressure systems. All of this came before and helped to inspire, the electric telegraphs that would -eventually - become the basis for the electric systems of communication we use today for the modern internet. So, to put a long story short, you better thank the Greeks and their system of sending codes long distances using containers and water for the modern communications that we get to enjoy today! developed

The Hydraulic Telegraph: The First Data Network All The Way From Ancient Greece!

Recently, while immersed in a game called "Factorio," I noticed striking parallels to the challenges encountered as the pioneer member of a data analytics team in a startup. In "Factorio," the game narrative revolves around an astronaut stranded on an alien planet with only raw materials at his disposal, tasked to build a spaceship for his escape. This scenario felt reminiscent of the real-world process of establishing an analytics team from scratch in a burgeoning startup. Let's visualize this scenario. You're just appointed the inaugural data and analytics team member in a fresh, promising startup. In your arsenal, you have an innovative product, abundant resources, and a horizon teeming with potential priorities. The questions then arise - where should you begin? What should be your initial thought process? How do you sequentially develop your strategy to construct an effective analytical team from scratch, which would eventually serve as a launchpad for the startup's trajectory? This article aims to delve into these intricate dynamics. : While this article draws parallels with the game Factorio, prior knowledge of the game isn't necessary to understand the insights shared; and for those familiar with Factorio, please pardon any oversimplification of the game mechanics for the sake of analogy. Disclaimer If you've ever wondered what it feels like to crash land on an alien planet with only your raw resources, congratulations – you're in a startup! Remember, a spaceship crash is just a pretty dramatic product pivot. Table of Contents Phase 1.  Manual/foundation Phase 2. Initiating automation Phase 3. Scaling Phase 4. Flight Conclusion As the inaugural member of the team in a startup, your journey is reminiscent of the first moments in "Factorio." Much like the astronaut, you're venturing into an alien planet, equipped with raw resources and the tools salvaged from your spaceship. These tools symbolize your wealth of knowledge, skills, and experience. Your mission? Transform these primary resources into a functioning rocket ship - a successful, scalable data analytics team. data analytics Phase 1. Manual/foundation In the initial phase of this mission, you're in exploration mode, much like the astronaut surveying his alien environment. You observe the data landscape, discover data sources (similar to mines of raw materials), and map out potential obstacles. This stage is all about understanding the topography of your new environment. It is a stage marked by high personal involvement, heightened curiosity, and hands-on work. You begin by manually gathering data from multiple sources, comparable to mining resources from different points in the game. You then create initial reports and analyses to better understand the product and business context. These reports serve as the first bricks in your rocket ship factory, offering valuable insights into your product's performance and customer behavior. Simultaneously, you forge relationships with various stakeholders in the startup - product managers, software engineers, marketers, and leadership. You communicate your findings, propose data-driven recommendations for product improvements, and even suggest simple A/B tests to validate your hypotheses. Your active involvement in these cross-functional dialogues enhances your understanding of the business and positions data as a crucial component in decision-making processes. Being the first data scientist in a startup is a bit like being the first one at the party - you're not quite sure where to put your coat, and you're a bit wary of double-dipping in the guacamole. However, this exploration and manual operation is challenging. This stage is labor-intensive and demands high levels of personal involvement. Analytics at this stage is manual, like a blacksmith forging tools by hand. Reporting involves connecting directly to data sources, stirring, and shaking data to understand what's happening. An ad-hoc analysis is frequent, which helps to understand unusual instances or random questions. Your nascent analytics infrastructure may also encounter hiccups that necessitate your manual intervention, sometimes leading to moments of realization about a forgotten piece that isn't working as it should. This stage can feel like a one-person show, where you're running around, juggling various tasks, and ensuring everything operate smoothly. While this phase may not be flexible or sustainable in the long run, it serves a vital purpose. It equips you with a deep understanding of the data landscape, the intricacies of the product, and the broader business context. You're intimately involved in shaping the foundation of your rocket ship factory, which imparts a rewarding sense of ownership and accomplishment. However, it's crucial to recognize when to transition from this manual, labor-intensive phase to a more automated, extendable model to avoid stagnation and burnout. This realization marks the end of stage one and the beginning of stage two: initiating automation. It's a vital turning point in your journey, keeping the evolution from a solitary explorer to a strategic architect aiming to build a robust, scalable data analytics team. Practical actions you can take at this stage Understand its sources, the relationships between different types of data, and the overall data architecture in your startup. Get to know your data. Start with developing essential reports that provide vital insights into your product or service. These can include user engagement metrics, performance metrics, sales, and revenue metrics, etc. Establish baseline reporting. These are custom, one-time pieces of research that are done on demand. Be prepared to perform ad-hoc analyses. Work closely with your colleagues, particularly those in product management and software engineering. Understand their needs, how they use data, and how you can help them make data-informed decisions. Collaborate with stakeholders. Phase 2. Initiating automation The initial stage's manual labor and high personal involvement serve a crucial purpose – they provide an intimate understanding of the data, business, and product context. This understanding is comparable to a detailed map of the alien planet in "Factorio," highlighting the rich mines of data, potential obstacles, and areas for exploration. However, a shift towards automation is necessary to ensure scalability and efficiency. As you transition into this phase, your role morphs from a hands-on miner to an architect. You begin to design processes that optimize the transformation of raw data into valuable insights, like setting up machines to automate the mining and refining of resources in the game. One of your primary tasks in this phase involves identifying bottlenecks in your processes – repeated frequently, essential for smooth operation, consuming a significant amount of your time, and, most importantly, can be automated. In the game of Factorio, it's akin to finding out which parts of the resource processing can be automated to free up time and boost efficiency. These tasks could range from ad-hoc analyses required for primary product metrics reporting, studies essential for A/B tests (such as power analysis for the new test and post-test analysis of the A/B test metrics), or even routine data gathering, joining, and transformation to answer recurring business questions. When setting up your initial data infrastructure, you'll often encounter a piece of code that no one dares to touch. It's like the support beam in a crumbling building - we don't know why it's holding everything up, but we're not about to pull it out to find out. As you identify these tasks, you adopt some of three primary strategies to streamline your workflow: : In "Factorio," a factory is an assembly line of automated machinery that takes in raw materials and churns out finished goods. In the world of data analytics, a 'factory' is a suite of tools and systems designed to automate your workflow. These may include workflow schedulers like Airflow, which automate regular tasks, dashboard systems such as Redash or Grafana, which automate reporting and foster a self-service environment for stakeholders; and ETL/ELT pipelines which automate data transportation and transformation. Building a factory : This is similar to inviting friends for a cooperative game in Factorio. Expanding your team involves hiring more data specialists. They bring diverse skill sets and perspectives, which accelerates progress and helps build robust analytics capabilities. Expanding your team : In Factorio, a technology tree represents all the possible upgrades and technological advancements. Similarly, in a startup, it refers to planning and implementing the necessary technology stack that supports your data needs, ensuring the scalable, long-term growth of your analytics operations. Developing a technology tree However, it's essential to maintain a balance during this phase. While it may be tempting to automate every process and hire extensively, ensuring these decisions align with the startup's financial resources and strategic goals is critical. Additionally, remember to design flexible systems that can adapt to evolving business needs – avoid creating a complicated, high-maintenance infrastructure that could become a burden in the future. The transition from stage one to stage two reminds me of evolving from an explorer to an architect in Factorio. It marks a significant shift in how you interact with data and sets the stage for your analytics team's scalable, efficient growth. The journey is far from over, and there are still stages to navigate, but the groundwork you've laid here will be instrumental in the coming stages. Practical actions you can take at this stage: This could include data cleaning tools, reporting software, dashboarding tools, etc. Look for tools that can help you automate your routine tasks. It can save you considerable time and increase efficiency. Identify tasks you perform frequently and look for ways to automate them. It involves setting up data pipelines, implementing a data warehouse, and setting up ETL processes. Start developing your data infrastructure. Look for individuals with skills that complement your own and can help you scale your operations. As you move to this stage, you may need to expand your team. Phase 3. Scaling In the previous stages, you've laid the foundation and initiated automation for your analytics operations, much like building a rudimentary factory on an alien planet in "Factorio." Now comes the time to scale, expand your operations, and prepare for growing demands. This phase is similar to increasing your factory, optimizing production lines, and preparing for larger, more complex challenges in the game. At this juncture, your analytics system is mainly automated, doing much of the routine work without your constant involvement. The system has morphed from a basic rocket ship factory into a more sophisticated operation. Yet, this is not the moment to rest but to gear up for more significant challenges. It's similar to the game scenario where resources become scarcer and demand multiplies - it calls for a strategic expansion. The scaling phase has three crucial components: : In "Factorio," as you exhaust local resources, you must build train networks to import materials from distant mines. Similarly, as your internal data resources need to be improved to drive deeper insights in a startup scenario, you might need to integrate third-party data or expertise. This could mean adopting external data APIs, forging partnerships for data sharing, or hiring external consultants or agencies with niche skills. External resource integration : Much like setting up dashboards to monitor your factory's efficiency in Factorio, it is crucial to establish robust performance tracking in a data analytics setup. Regularly monitoring your analytics system's performance, peak capacity, and identification of bottlenecks ensure its smooth operation. It also provides insights for future improvements and helps in proactive troubleshooting. Monitoring : In "Factorio," you often have to restructure your factory to accommodate new technologies or to make the production line more efficient. Similarly, in a data environment, refactoring or re-engineering your data pipelines, models, or reports might be necessary to improve efficiency, reduce computational costs, or incorporate changes in the business context. Refactoring Throughout this stage, the critical emphasis is on efficiency and optimization. While aiming for massive growth is tempting, the scale phase is more about 'smart' growth. It's about ensuring that your operations are extendable, maintainable, and flexible enough to adapt to the rapid changes characteristic of startup environments. Scaling up your processes is a bit like building your IKEA furniture - you start off with manual labor, realize you need some automation (or at least a power drill), and eventually, you’re standing back admiring your work, just to realize you’ve got some leftover parts. Did we really need that data pipeline, or was it just an optional extra? Scaling also includes planning for future growth. While you want to ensure that your current setup is efficient, you also want to anticipate the needs of your growing organization. This might mean investing in scalable technologies, planning for increased data storage and processing capabilities, or even initiating discussions about building a data warehouse. Much like the game "Factorio," the challenges are becoming more complex and demanding, but the opportunities for creativity and impact are also growing. As you navigate this stage, your role evolves from an architect to a city planner, meticulously planning and overseeing the growth and development of your analytics operations. And as you do so, you lay the groundwork for the final stage - the flight. Practical actions you can take at this stage Implement tools and processes to monitor the performance of your data infrastructure. Identify bottlenecks and areas for improvement. Continually look for ways to optimize your processes and data infrastructure. As your operations grow, you may find that specific processes or systems no longer meet your needs. Be prepared to refactor or redesign these as necessary. You may need to start integrating external data sources or systems. Establish new data pipelines or expand your data infrastructure. Phase 4. Flight Much like in "Factorio," where the ultimate goal is to build a rocket and escape the alien planet, in the context of a startup's analytics team, the 'flight' phase represents reaching a stage where your analytics operations are fully-functional, robust, flexible, and are driving strategic decisions and value creation. At this juncture, you're not just managing a data analytics operation; you're overseeing a well-oiled machine contributing significantly to the startup's trajectory. You have built a system that functions almost independently, only requiring your attention for high-level strategic decisions and occasional troubleshooting. However, just as in the game, reaching the final stage doesn't mean you're free from challenges. Even at this level, there are hurdles to overcome: : In "Factorio," the aliens evolve and become more of a threat as you progress. Similarly, competitors will continue to grow and innovate in a startup environment. You must ensure that your analytical capabilities continue to develop, staying ahead of the curve and providing the strategic edge your startup needs to thrive. Evolving competition : Time is a critical resource, especially for startups. Despite having a successful operation, you may need more help. It's essential to ensure that your analytics operation runs efficiently and that resources are allocated strategically to support continual growth and innovation. Resource constraints : Over time, some of your data processes and models may lose effectiveness or become outdated. Just as you would upgrade your factories and technologies in "Factorio," it's essential to regularly revisit and improve your analytics operations. Keeping your analytics tools and techniques up-to-date ensures you continue generating valuable insights. Decreasing marginal utility Therefore, the 'flight' phase is not about coasting but continual improvement, adaptation, and innovation. It's about maintaining the momentum built up and pushing for greater heights. Practical actions you can take at this stage Continually look for ways to improve your operations, whether through new tools, processes, or data analysis approaches. Don't rest on your laurels. This might involve learning new tools or techniques, attending conferences, or reading relevant publications. Keep up with developments in the field of data science. Use your data insights to inform these decisions and drive growth. You should contribute to your startup's strategic decisions at this stage. . Ensure that you are allocating your time, your team's time, and your budget to maximize return on investment. Be mindful of your resources Conclusion Establishing and growing an analytics team in a 0 → 1 startup environment can be an exciting, challenging, and rewarding journey - much like progressing through the stages of the game. Starting from basic manual work, you gradually build your operations, initiate automation, scale your capabilities, and finally reach the flight stage, where your analytics operations become vital to your startup's journey. However, each step requires different skills, strategies, and perspectives. Like in the game, you need to continually adapt, innovate, and strategize to ensure the success of your analytics operations and, consequently, your startup. By understanding and implementing this four-stage approach, you can survive and thrive in a 0 → 1 analytics team.

From Crashing to Lift-Off: How to Thrive as the First Data Scientist in a Startup

ক্র্যাশিং থেকে লিফ্ট-অফ পর্যন্ত: একটি স্টার্টআপে প্রথম ডেটা সায়েন্টিস্ট হিসাবে কীভাবে উন্নতি করবেন

Vom Absturz zum Start: Wie Sie als erster Datenwissenschaftler in einem Startup erfolgreich sind

충돌에서 이륙까지: 스타트업에서 최초의 데이터 과학자로 성공하는 방법

Çökmeden Kalkışa: Bir Startup'taki İlk Veri Bilimcisi Olarak Nasıl Başarılı Olunur?

You’re a founder. You know you need to have a “data play” (or worse, “AI play”). Investors and clients are asking about machine (or worse, deep learning). The question is no longer why, but when. startup learning So, you hire your first data scientist. A fresh graduate (those with industry experience are too expensive and hard to recruit). They have a PhD (so you can nonchalantly drop ‘I have my best PhDs looking into this’ in the conversation with potential clients and VCs.) They‘re excited about doing machine learning — and so are you. This is going to rock! Your data scientist starts, ready to take your data and build models from it, just like you agreed. Back in academia, they were working with datasets that several generations of grad students have been using. They’re not naive, though — they know this time around they’ll have to create their own. No problem! Your startup has a year worth of logs in S3, ready to be mined and used as training data. How not to think about the data science Venn diagram There’s no infrastructure for data analysis, so everything has to be done from scratch. Your data scientist is trying to install various tools incompatible with your stack or architecture and, for now, is doing one-off log parsing to answer every question. The engineering team doesn’t feel it’s safe to give them prod access so they provide an offline copy of your database. you say to the data scientist. There! You have access to all the data you need! Except… data in that offline database is, by design, not structured in a way that’s easy to analyze or combine with the data in the logs. Values don’t make sense or are missing for weeks at a time, the only events instrumented are those relevant to the ops team, and even the simplest query takes forever to run. Everybody is frustrated . The data scientist wanted to work on machine learning — those elegant algorithms they spent their career thinking and publishing about. Yes, they expected to put some effort into gathering and cleaning data — but they didn’t expect it to be so complex and messy, with so much missing or hard to access. They didn’t expect to spend so much time in meetings or Slack with questions about how the data was gathered and what, exactly, is in that json blob. They didn’t expect the rest of the company to care so little about how each redesign messes up month to month comparisons. They didn’t realize nobody logged all the options a user saw before they clicked, nor the version of the UI at the time, so that one year of training data is not exactly usable. You hired a data scientist to work on machine learning. They’ve been warned about spending 80% of their time cleaning data. That sounds like a distant dream now. Instead, they spend 80% of their time begging for data to be created, accessed, moved, or explained. They’re spending the other 20% lobbying for data science-friendly tools, security policies, and infrastructure, or looking for a new job. The engineering team is frustrated, too. They have to take time away from their own work to do a ton of thankless grunt work for the scientist who sits around, does nothing useful and still manages to constantly complain about the data not being good enough. You, the founder, are even more frustrated. It’s been 2 months, and the data scientist didn’t even produce a decent dashboard, much less the magic machine learning everybody keeps raving about. Plus, they seem like a bad culture fit, slowing everybody else down for no reason. Deep down, you start suspecting machine learning is snake oil and . AI is just hype You keep this to yourself, though, and tell your investors and clients about your AI play and your crazy smart PhD data scientist who’s super excited to work on machine learning any day now. If you’re wondering how to avoid this painful but too frequent scenario, here’s the Quora answer that inspired this post : “ What are the challenges of building a data team at a startup ?”. Think other people would benefit from this article? Click the heart to help them find it ❤

How not to hire your first data scientist

Authors:
(1) Hanoona Rasheed, Mohamed bin Zayed University of AI and equally contributing first authors;
(2) Muhammad Maaz, Mohamed bin Zayed University of AI and equally contributing first authors;
(3) Sahal Shaji, Mohamed bin Zayed University of AI;
(4) Abdelrahman Shaker, Mohamed bin Zayed University of AI;
(5) Salman Khan, Mohamed bin Zayed University of AI and Australian National University;
(6) Hisham Cholakkal, Mohamed bin Zayed University of AI;
(7) Rao M. Anwer, Mohamed bin Zayed University of AI and Aalto University;
(8) Eric Xing, Mohamed bin Zayed University of AI and Carnegie Mellon University;
(9) Ming-Hsuan Yang, University of California - Merced and Google Research;
(10) Fahad S. Khan, Mohamed bin Zayed University of AI and Linköping University. Editor's Note: This is Part 4 of 10 of a study detailing the development of an AI model that is designed to describe images to users. Read the rest below. Table of Links Abstract and 1 Introduction
2. Related Work
3. Method
4. Data Annotation Pipeline
5. Experiments
6. Conclusion and References Supplementary Material (Part 1) A. Additional Implementation Details
B. Additional Downstream Tasks
C. Additional Qualitative Results Supplementary Material (Part 2) D. Dataset Visualization
E. Limitations and Future Work
F. Ethics and Societal Impact 4. Data Annotation Pipeline We introduce our automated annotation pipeline used to create the Grounding-anything Dataset (GranD). GranD is a comprehensive, multi-purpose image-text dataset offering a range of contextual information, from fine-grained to high-level details. It aims to overcome challenges in image understanding and dense pixel-level grounding, thereby expanding the capabilities of visual instruction tuning in LMMs. The pipeline contains four distinct levels (see Fig. 4). i) Level-1 focuses on object localization and provides semantic labels, segmentation masks, attributes, and depth information. ii) Level-2 defines relationships between detected objects. iii) Level-3 organizes information from the first two levels into a hierarchical scene graph, used to generate dense captions using LLM with in-context examples. iv) Level-4 offers enriched contextual information for a deeper understanding of the scene, going beyond what’s observed (e.g., historical information of a landmark). Please refer to Appendix A.4 for pipeline implementation details. 4.1. Object Localization and Attributes (Level-1) In level-1, the focus is on detailed object identification within images. First, object-bounding boxes are identified using multiple SoTA object detection models. Classagnostic NMS is applied to each model to filter out false positives. After this step, bounding boxes from different models are compared using IoU, with a bounding box retained as an object only if detected by at least two other detection models. We also generate attributes for each filtered object using region-based vision-language models and incorporate depth information to contextualize each object’s relative position within the scene. 4.2. Relationships and Landmarks (Level-2) In level-2, multiple short textual descriptions of the overall scene are generated. Phrases extracted from these descriptions are grounded to specific objects in level-1 to form relationships. These relationships articulate connections between multiple objects or define an object’s role within the scene. Further, each scene is assigned a landmark category that includes a primary and a more specific sub-category (see Tab. 7 in Appendix 7). 4.3. Scene Graph and Dense Captioning (Level-3) In level-3, object attributes and labels from level-1 are combined with the relationships and phrases obtained from level-2 to form a hierarchical scene graph. This structured data serves as a query for LLM to generate dense image captions. To provide additional context, depth values and bounding box coordinates are used to assign each object to specific spatial layers within the scene, such as immediate foreground, foreground, midground, or background. Additionally, short scene-level captions are incorporated into the scene graph to enhance LLMs’ contextual understanding. Dense Captioning Verification: To enhance the fidelity of the LLM-generated dense captions, we implement an automatic verification pipeline using chain-of-thoughts prompting. This pipeline produces a checklist of objects derived from the generated dense caption expected to be present in the image. The associated caption is flagged as inaccurate if any object specified in the checklist is absent from the scene graph. Such captions are then regenerated, incorporating feedback from the initial assessment. 4.4. Extra Contextual Insights (Level-4) Level-4 builds on the scene graph from level-3 to obtain a more detailed visual understanding. we query LLM to extract extended contextual insights beyond basic object identification and relationships, including details about the landmarks, historical context, guidelines for interacting with the scene, and even predictive elements about future events. To facilitate this, we prompt LLM with in-context examples. Utilizing our automated annotation pipeline, we annotate a corpus of 11M SAM images [18], which are inherently diverse, high-resolution, and privacy-compliant. The resulting dataset comprises 810M regions, each associated with a segmentation mask, and includes 7.5M unique concepts. Further, the dataset features 84M referring expressions, 22M grounded short captions, and 11M densely grounded captions. To our knowledge, this is the first dataset of this scale generated entirely through an automated annotation pipeline (see Tab. 2 for details and Fig. 15 for dataset sample visualizations). We extend open-source datasets—namely Flickr30K [37], RefCOCOg [16], and PSG [49] by generating compatible GCG annotations. For RefCOCOg, we use the dataset’s referring expressions and their connected masks. These expressions offer concise descriptions of distinct objects in the image. With the aid of GPT-4, we seamlessly blend these referring expressions with contextual information from COCO captions, crafting detailed yet accurate grounded captions while preserving the original referring expressions. This ensures zero error in matching phrases with their corresponding segmentation masks. This technique yields approximately 24K GCG samples. For PSG, we leverage the dataset’s triplet structures, which describe relations between two objects in a scene. These triplets are integrated with COCO captions using GPT-4, resulting in densely annotated captions that can be mapped to segmentation masks. This gives us around 31K additional GCG samples. For Flickr-30K, we use the 158K Flickr captions and their referring expressions alongside associated bounding boxes. These boxes are then accurately segmented using HQ-SAM [17]. In addition, we contribute a minor, high-quality manual annotation set to benchmark the GCG task. Using GranD’s automatic annotations as a base, annotators refine referring expressions to match SAM GT masks, yielding around 1000 focused samples for training and 1000 for evaluation (refer to Appendix D and Fig. 14 for designed prompts and dataset visualizations). This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Hanoona Rasheed, Mohamed bin Zayed University of AI and equally contributing first authors; (2) Muhammad Maaz, Mohamed bin Zayed University of AI and equally contributing first authors; (3) Sahal Shaji, Mohamed bin Zayed University of AI; (4) Abdelrahman Shaker, Mohamed bin Zayed University of AI; (5) Salman Khan, Mohamed bin Zayed University of AI and Australian National University; (6) Hisham Cholakkal, Mohamed bin Zayed University of AI; (7) Rao M. Anwer, Mohamed bin Zayed University of AI and Aalto University; (8) Eric Xing, Mohamed bin Zayed University of AI and Carnegie Mellon University; (9) Ming-Hsuan Yang, University of California - Merced and Google Research; (10) Fahad S. Khan, Mohamed bin Zayed University of AI and Linköping University. Authors: Authors: (1) Hanoona Rasheed, Mohamed bin Zayed University of AI and equally contributing first authors; (2) Muhammad Maaz, Mohamed bin Zayed University of AI and equally contributing first authors; (3) Sahal Shaji, Mohamed bin Zayed University of AI; (4) Abdelrahman Shaker, Mohamed bin Zayed University of AI; (5) Salman Khan, Mohamed bin Zayed University of AI and Australian National University; (6) Hisham Cholakkal, Mohamed bin Zayed University of AI; (7) Rao M. Anwer, Mohamed bin Zayed University of AI and Aalto University; (8) Eric Xing, Mohamed bin Zayed University of AI and Carnegie Mellon University; (9) Ming-Hsuan Yang, University of California - Merced and Google Research; (10) Fahad S. Khan, Mohamed bin Zayed University of AI and Linköping University. Editor's Note: This is Part 4 of 10 of a study detailing the development of an AI model that is designed to describe images to users. Read the rest below. Editor's Note: This is Part 4 of 10 of a study detailing the development of an AI model that is designed to describe images to users. Read the rest below. Editor's Note: This is Part 4 of 10 of a study detailing the development of an AI model that is designed to describe images to users. Read the rest below. Editor's Note: This is Part 4 of 10 of a study detailing the development of an AI model that is designed to describe images to users. Read the rest below. Table of Links Abstract and 1 Introduction 2. Related Work 3. Method 4. Data Annotation Pipeline 5. Experiments 6. Conclusion and References Abstract and 1 Introduction Abstract and 1 Introduction 2. Related Work 2. Related Work 3. Method 3. Method 4. Data Annotation Pipeline 4. Data Annotation Pipeline 5. Experiments 5. Experiments 6. Conclusion and References 6. Conclusion and References Supplementary Material (Part 1) Supplementary Material (Part 1) A. Additional Implementation Details B. Additional Downstream Tasks C. Additional Qualitative Results A. Additional Implementation Details A. Additional Implementation Details B. Additional Downstream Tasks B. Additional Downstream Tasks C. Additional Qualitative Results C. Additional Qualitative Results Supplementary Material (Part 2) Supplementary Material (Part 2) D. Dataset Visualization E. Limitations and Future Work F. Ethics and Societal Impact D. Dataset Visualization D. Dataset Visualization E. Limitations and Future Work E. Limitations and Future Work F. Ethics and Societal Impact F. Ethics and Societal Impact 4. Data Annotation Pipeline We introduce our automated annotation pipeline used to create the Grounding-anything Dataset (GranD). GranD is a comprehensive, multi-purpose image-text dataset offering a range of contextual information, from fine-grained to high-level details. It aims to overcome challenges in image understanding and dense pixel-level grounding, thereby expanding the capabilities of visual instruction tuning in LMMs. The pipeline contains four distinct levels (see Fig. 4). i) Level-1 focuses on object localization and provides semantic labels, segmentation masks, attributes, and depth information. ii) Level-2 defines relationships between detected objects. iii) Level-3 organizes information from the first two levels into a hierarchical scene graph, used to generate dense captions using LLM with in-context examples. iv) Level-4 offers enriched contextual information for a deeper understanding of the scene, going beyond what’s observed (e.g., historical information of a landmark). Please refer to Appendix A.4 for pipeline implementation details. 4.1. Object Localization and Attributes (Level-1) In level-1, the focus is on detailed object identification within images. First, object-bounding boxes are identified using multiple SoTA object detection models. Classagnostic NMS is applied to each model to filter out false positives. After this step, bounding boxes from different models are compared using IoU, with a bounding box retained as an object only if detected by at least two other detection models. We also generate attributes for each filtered object using region-based vision-language models and incorporate depth information to contextualize each object’s relative position within the scene. 4.2. Relationships and Landmarks (Level-2) In level-2, multiple short textual descriptions of the overall scene are generated. Phrases extracted from these descriptions are grounded to specific objects in level-1 to form relationships. These relationships articulate connections between multiple objects or define an object’s role within the scene. Further, each scene is assigned a landmark category that includes a primary and a more specific sub-category (see Tab. 7 in Appendix 7). 4.3. Scene Graph and Dense Captioning (Level-3) In level-3, object attributes and labels from level-1 are combined with the relationships and phrases obtained from level-2 to form a hierarchical scene graph. This structured data serves as a query for LLM to generate dense image captions. To provide additional context, depth values and bounding box coordinates are used to assign each object to specific spatial layers within the scene, such as immediate foreground, foreground, midground, or background. Additionally, short scene-level captions are incorporated into the scene graph to enhance LLMs’ contextual understanding. Dense Captioning Verification: To enhance the fidelity of the LLM-generated dense captions, we implement an automatic verification pipeline using chain-of-thoughts prompting. This pipeline produces a checklist of objects derived from the generated dense caption expected to be present in the image. The associated caption is flagged as inaccurate if any object specified in the checklist is absent from the scene graph. Such captions are then regenerated, incorporating feedback from the initial assessment. Dense Captioning Verification: 4.4. Extra Contextual Insights (Level-4) Level-4 builds on the scene graph from level-3 to obtain a more detailed visual understanding. we query LLM to extract extended contextual insights beyond basic object identification and relationships, including details about the landmarks, historical context, guidelines for interacting with the scene, and even predictive elements about future events. To facilitate this, we prompt LLM with in-context examples. Utilizing our automated annotation pipeline, we annotate a corpus of 11M SAM images [18], which are inherently diverse, high-resolution, and privacy-compliant. The resulting dataset comprises 810M regions, each associated with a segmentation mask, and includes 7.5M unique concepts. Further, the dataset features 84M referring expressions, 22M grounded short captions, and 11M densely grounded captions. To our knowledge, this is the first dataset of this scale generated entirely through an automated annotation pipeline (see Tab. 2 for details and Fig. 15 for dataset sample visualizations). We extend open-source datasets—namely Flickr30K [37], RefCOCOg [16], and PSG [49] by generating compatible GCG annotations. For RefCOCOg, we use the dataset’s referring expressions and their connected masks. These expressions offer concise descriptions of distinct objects in the image. With the aid of GPT-4, we seamlessly blend these referring expressions with contextual information from COCO captions, crafting detailed yet accurate grounded captions while preserving the original referring expressions. This ensures zero error in matching phrases with their corresponding segmentation masks. This technique yields approximately 24K GCG samples. For PSG, we leverage the dataset’s triplet structures, which describe relations between two objects in a scene. These triplets are integrated with COCO captions using GPT-4, resulting in densely annotated captions that can be mapped to segmentation masks. This gives us around 31K additional GCG samples. For Flickr-30K, we use the 158K Flickr captions and their referring expressions alongside associated bounding boxes. These boxes are then accurately segmented using HQ-SAM [17]. In addition, we contribute a minor, high-quality manual annotation set to benchmark the GCG task. Using GranD’s automatic annotations as a base, annotators refine referring expressions to match SAM GT masks, yielding around 1000 focused samples for training and 1000 for evaluation (refer to Appendix D and Fig. 14 for designed prompts and dataset visualizations). This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

Researchers Develop Groundbreaking Method to Teach AI to 'Understand' Images Like Never Before

Authors:
(1) Guillaume Staerman, INRIA, CEA, Univ. Paris-Saclay, France;
(2) Marta Campi, CERIAH, Institut de l’Audition, Institut Pasteur, France;
(3) Gareth W. Peters, Department of Statistics & Applied Probability, University of California Santa Barbara, USA. Table of Links Abstract and 1. Introduction 2. Background & Preliminaries 2.1. Functional Isolation Forest 2.2. The Signature Method 3. Signature Isolation Forest Method 4. Numerical Experiments 4.1. Parameters Sensitivity Analysis 4.2. Advantages of (K-)SIF over FIF 4.3. Real-data Anomaly Detection Benchmark 5. Discussion & Conclusion, Impact Statements, and References Appendix A. Additional Information About the Signature B. K-SIF and SIF Algorithms C. Additional Numerical Experiments 4.1. Parameters Sensitivity Analysis We investigate the behavior of K-SIF and SIF with respect to their two main parameters: the depth of the signature k and the number of split windows ω. For the sake of place, the experiment on the depth is postponed in Section C.1 in the Appendix. The Role of the Signature Split Window. The number of split windows allows the extraction of information over specific intervals (randomly selected) of the underlying data. Thus, at each tree node, the focus will be on a particular portion of the data, which is the same across all the sample curves for comparison purposes. This approach ensures that the analysis is performed on comparable sections of the data, providing a systematic way to examine and compare different intervals or features across the sample curves. We explore the role of this parameter with two different datasets that reproduce two types of anomaly scenarios. The first considers isolated anomalies in a small interval, while the second contains persistent ones across all the function parametrization. In this way, we observe the behavior of K-SIF and SIF with respect to different types of anomalies. The first dataset is constructed as follows. We simulate 100 constant functions. We then select at random 90% of these curves and Gaussian noise on a sub-interval; for the remaining 10% of the curves, we add Gaussian noise on another sub-interval, different from the first one. More precisely: • 90% of the curves, considered as normal, are generated according to with ε(t) ∼ N (0, 1), b ∼ U([0, 100]) and U representing the uniform distribution. • 10% of the curves, considered as abnormal, are generated according to where ε(t) ∼ N (0, 1) and b ∼ U([0, 100]). We simulate at random 90% of the paths with µ = 0, σ = 0.5, and consider them as normal data. Then, the remaining 10% are simulated with drift µ = 0.2, standard deviation σ = 0.4, and considered abnormal data. We compute K-SIF with different numbers of split windows, varying from 1 to 10, with a truncation level set equal to 2 and N = 1, 000 the number of trees. The experiment is repeated 100 times, and we report the averaged AUC under the ROC curves in Figure 1 for both datasets and three pre-selected dictionaries. For the first dataset, where anomalies manifest in a small portion of the functions, increasing the number of splits significantly enhances the algorithm’s performance in detecting anomalies. The performance improvement shows a plateau after nine split windows. In the case of the second dataset with persistent anomalies, a higher number of split windows has a marginal impact on the algorithm’s performance, maintaining satisfactory results. Therefore, without prior knowledge about the data, opting for a relatively high number of split windows, such as 10, would ensure robust performance in both scenarios. Additionally, a more significant number of split windows enables the computation of the signature on a smaller portion of the functions, leading to improved computational efficiency. This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Guillaume Staerman, INRIA, CEA, Univ. Paris-Saclay, France; (2) Marta Campi, CERIAH, Institut de l’Audition, Institut Pasteur, France; (3) Gareth W. Peters, Department of Statistics & Applied Probability, University of California Santa Barbara, USA. Authors: Authors: (1) Guillaume Staerman, INRIA, CEA, Univ. Paris-Saclay, France; (2) Marta Campi, CERIAH, Institut de l’Audition, Institut Pasteur, France; (3) Gareth W. Peters, Department of Statistics & Applied Probability, University of California Santa Barbara, USA. Table of Links Abstract and 1. Introduction Abstract and 1. Introduction 2. Background & Preliminaries 2. Background & Preliminaries 2.1. Functional Isolation Forest 2.1. Functional Isolation Forest 2.2. The Signature Method 2.2. The Signature Method 3. Signature Isolation Forest Method 3. Signature Isolation Forest Method 4. Numerical Experiments 4. Numerical Experiments 4.1. Parameters Sensitivity Analysis 4.1. Parameters Sensitivity Analysis 4.2. Advantages of (K-)SIF over FIF 4.2. Advantages of (K-)SIF over FIF 4.3. Real-data Anomaly Detection Benchmark 4.3. Real-data Anomaly Detection Benchmark 5. Discussion & Conclusion, Impact Statements, and References 5. Discussion & Conclusion, Impact Statements, and References Appendix A. Additional Information About the Signature A. Additional Information About the Signature B. K-SIF and SIF Algorithms B. K-SIF and SIF Algorithms C. Additional Numerical Experiments C. Additional Numerical Experiments 4.1. Parameters Sensitivity Analysis We investigate the behavior of K-SIF and SIF with respect to their two main parameters: the depth of the signature k and the number of split windows ω. For the sake of place, the experiment on the depth is postponed in Section C.1 in the Appendix. The Role of the Signature Split Window. The number of split windows allows the extraction of information over specific intervals (randomly selected) of the underlying data. Thus, at each tree node, the focus will be on a particular portion of the data, which is the same across all the sample curves for comparison purposes. This approach ensures that the analysis is performed on comparable sections of the data, providing a systematic way to examine and compare different intervals or features across the sample curves. The Role of the Signature Split Window. We explore the role of this parameter with two different datasets that reproduce two types of anomaly scenarios. The first considers isolated anomalies in a small interval, while the second contains persistent ones across all the function parametrization. In this way, we observe the behavior of K-SIF and SIF with respect to different types of anomalies. The first dataset is constructed as follows. We simulate 100 constant functions. We then select at random 90% of these curves and Gaussian noise on a sub-interval; for the remaining 10% of the curves, we add Gaussian noise on another sub-interval, different from the first one. More precisely: • 90% of the curves, considered as normal, are generated according to with ε(t) ∼ N (0, 1), b ∼ U([0, 100]) and U representing the uniform distribution. • 10% of the curves, considered as abnormal, are generated according to where ε(t) ∼ N (0, 1) and b ∼ U([0, 100]). We simulate at random 90% of the paths with µ = 0, σ = 0.5, and consider them as normal data. Then, the remaining 10% are simulated with drift µ = 0.2, standard deviation σ = 0.4, and considered abnormal data. We compute K-SIF with different numbers of split windows, varying from 1 to 10, with a truncation level set equal to 2 and N = 1, 000 the number of trees. The experiment is repeated 100 times, and we report the averaged AUC under the ROC curves in Figure 1 for both datasets and three pre-selected dictionaries. For the first dataset, where anomalies manifest in a small portion of the functions, increasing the number of splits significantly enhances the algorithm’s performance in detecting anomalies. The performance improvement shows a plateau after nine split windows. In the case of the second dataset with persistent anomalies, a higher number of split windows has a marginal impact on the algorithm’s performance, maintaining satisfactory results. Therefore, without prior knowledge about the data, opting for a relatively high number of split windows, such as 10, would ensure robust performance in both scenarios. Additionally, a more significant number of split windows enables the computation of the signature on a smaller portion of the functions, leading to improved computational efficiency. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

Decoding Split Window Sensitivity in Signature Isolation Forests

Afkodning af delt vinduesfølsomhed i signaturisolationsskove

Dekodierung der Split-Window-Empfindlichkeit in Signaturisolationswäldern

Descifrando la sensibilidad de la ventana dividida en bosques de aislamiento de firmas

Az osztott ablak érzékenységének dekódolása a Signature Isolation Forestsben

Decodifica della sensibilità della finestra divisa nelle foreste di isolamento delle firme

署名分離フォレストにおける分割ウィンドウの感度のデコード

Decoding Split Window Sensitivity ka Dithokgweng tša go Ikarola tša Mosaeno

Декодирование чувствительности разделенного окна в лесах изоляции сигнатур

Decoding Gutandukanya Idirishya Sensitivity mumashyamba yo kwigunga

Kusimbua Unyeti wa Dirisha la Kugawanyika katika Misitu ya Kutengwa kwa Sahihi

دستخطی تنہائی کے جنگلات میں اسپلٹ ونڈو کی حساسیت کو ڈی کوڈنگ کرنا

Ukuqonda Ukwahlulwa Kwefestile Yeefestile KuMsayino Wokwahlulwa Kwamahlathi

Do you read confidentiality agreements? Few people waste time. They all look alike, are boring, and probably don't lie. Companies, especially big ones, try to be honest about their responsibilities regarding user data if only to avoid hefty fines. And yet, something is wrong with the world's privacy protections. What if the concept of privacy itself is the problem? Decentralized projects don't usually say “privacy”; rather, they say data is “owned by the user” (data ownership). What's the difference? I came up with a fun analogy and want to share it. Let's take a more understandable subject - land, for example. If you own a plot of land, you have the right to get there whenever you want, build a house on it, make money with it, invite your friends to a party, and forbid anyone to enter. Ownership is the right to make decisions about some object. From a legal point of view, it consists of three elements: Possession (i.e., the possibility of unrestricted access)


Use (i.e., benefiting from the property)


Disposition (the right to sell, gift, or destroy the object) An owner can transfer some of his or her rights to another. For example, if you lease land, you will allow the tenant access and use, but not disposal. This is the classic, universal understanding of ownership. There are good reasons to extend it to include information. First, data plays a role in the modern economy; in the 21st century, it has become a valuable asset with measurable value. Look at the largest companies in the world: 4 out of 10 derive a significant part of their revenues from user data. Second, thanks to decentralized architecture and asymmetric cryptography, this has become feasible in principle: data ownership can be enforced. “Privacy” means that the data is kept secret and protected by the server. The server takes over the authentication and authorization functions to make the protection work. At its discretion, it can allow or deny someone access and use. In other words, the server owns the decisions - and so, it turns out, does the data. Can the server block a user? It can. Can it grant access to someone else for some reason - like a hacker? It can. Can someone else use your data for profit? Yes, that's the whole point of free digital services. The provider can destroy your account at will, as Google does, or take away your name, as Twitter does. And users are used to this state of affairs. As the former owner of the Music account on the service formerly known as Twitter put it, “Ultimately, they own the account.” As strange as it may sound, the idea of privacy on the Internet is reminiscent of medieval feudalism. In the Middle Ages, land was the main economic resource. Peasants used the land, but it belonged to the aristocracy. In return, the feudal lord was obligated to protect that land. Sound familiar? By allowing internet services to be in charge of privacy, we are adopting a feudal model: when our digital identity is protected, it is at the mercy of someone else. That's why tech companies cling so tightly to privacy. And why privacy laws being passed around the world exacerbate digital inequality. Internet giants limit our access to our own data, calling it an advantage. The very concept of privacy masks the fact that data does not belong to us. Let's be clear: this is not a global conspiracy theory but rather an attempt to find the root of the problem. Here's another classic legal principle: Cuius est solum, eius est usque ad coelum et ad inferos - To whomsoever it belongs, it is his all the way to the heavens (and all the way to hell) (Latin) I like the sound of that. This is exactly the kind of complete and unconditional ownership of one's own information made possible by decentralized, trustless networks with strong cryptography. That's why I believe Web3 is more than cryptocurrency speculation. It is the beginning of a new era of people's relationship with technology. Do you think the analogy with the Middle Ages is clickbait? Or does it seem to be true? Do you read confidentiality agreements? Few people waste time. They all look alike, are boring, and probably don't lie. Companies, especially big ones, try to be honest about their responsibilities regarding user data if only to avoid hefty fines. And yet, something is wrong with the world's privacy protections. What if the concept of privacy itself is the problem? Decentralized projects don't usually say “privacy”; rather, they say data is “owned by the user” (data ownership). What's the difference? I came up with a fun analogy and want to share it. Let's take a more understandable subject - land, for example. If you own a plot of land, you have the right to get there whenever you want, build a house on it, make money with it, invite your friends to a party, and forbid anyone to enter. Ownership is the right to make decisions about some object. From a legal point of view , it consists of three elements: From a legal point of view Possession (i.e., the possibility of unrestricted access) Use (i.e., benefiting from the property) Disposition (the right to sell, gift, or destroy the object) Possession (i.e., the possibility of unrestricted access) Possession (i.e., the possibility of unrestricted access) Use (i.e., benefiting from the property) Use (i.e., benefiting from the property) Disposition (the right to sell, gift, or destroy the object) Disposition (the right to sell, gift, or destroy the object) An owner can transfer some of his or her rights to another. For example, if you lease land, you will allow the tenant access and use, but not disposal. This is the classic, universal understanding of ownership. There are good reasons to extend it to include information. First, data plays a role in the modern economy; in the 21st century, it has become a valuable asset with measurable value. Look at the largest companies in the world : 4 out of 10 derive a significant part of their revenues from user data. Second, thanks to decentralized architecture and asymmetric cryptography, this has become feasible in principle: data ownership can be enforced. largest companies in the world “Privacy” means that the data is kept secret and protected by the server. The server takes over the authentication and authorization functions to make the protection work. At its discretion, it can allow or deny someone access and use. In other words, the server owns the decisions - and so, it turns out, does the data. Can the server block a user? It can. Can the server block a user? It can. Can it grant access to someone else for some reason - like a hacker? It can. Can it grant access to someone else for some reason - like a hacker? It can. Can someone else use your data for profit? Yes, that's the whole point of free digital services. Can someone else use your data for profit? Yes, that's the whole point of free digital services . whole point of free digital services The provider can destroy your account at will, as Google does , or take away your name, as Twitter does . Google does Twitter does And users are used to this state of affairs. As the former owner of the Music account on the service formerly known as Twitter put it, “Ultimately, they own the account.” As strange as it may sound, the idea of privacy on the Internet is reminiscent of medieval feudalism. In the Middle Ages, land was the main economic resource. Peasants used the land, but it belonged to the aristocracy. In return, the feudal lord was obligated to protect that land. Sound familiar? By allowing internet services to be in charge of privacy, we are adopting a feudal model: when our digital identity is protected, it is at the mercy of someone else. That's why tech companies cling so tightly to privacy. And why privacy laws being passed around the world exacerbate digital inequality. Internet giants limit our access to our own data, calling it an advantage. The very concept of privacy masks the fact that data does not belong to us. Let's be clear: this is not a global conspiracy theory but rather an attempt to find the root of the problem. Here's another classic legal principle: Cuius est solum, eius est usque ad coelum et ad inferos - To whomsoever it belongs, it is his all the way to the heavens (and all the way to hell) (Latin) Cuius est solum, eius est usque ad coelum et ad inferos - To whomsoever it belongs, it is his all the way to the heavens (and all the way to hell) (Latin) I like the sound of that. This is exactly the kind of complete and unconditional ownership of one's own information made possible by decentralized, trustless networks with strong cryptography. That's why I believe Web3 is more than cryptocurrency speculation. It is the beginning of a new era of people's relationship with technology. Do you think the analogy with the Middle Ages is clickbait? Or does it seem to be true?

FIRST DATA

firstdata.com

10K+ employees

Since 1984

COMPANY RANKING

FIRST DATA

EVERGREEN INDEX #5212

Representing this company? Click here to claim and customize this page!

First Data WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!

FIRST DATA

firstdata.com

10K+ employees

Since 1984

COMPANY RANKING

FIRST DATA

EVERGREEN INDEX #5212

RELATED COMPANIESRANK

STORIES ABOUT FIRST DATA

Representing this company? Click here to claim and customize this page!

First Data WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!