Alex 'Sandy' Pentland

#Resilience, the capacity to sustain core functions despite failures, is essential in #complex #networks. Most analyses rely on fixed node dynamics and simplified network structures, limiting their #realworld #applicability. Jianxi Gao from Rensselaer Polytechnic Institute, Yong Li from Tsinghua University and colleagues propose ResInf, a #deeplearning framework that combines transformers and graph #neural #networks to infer resilience directly from observational data. ResInf learns node dynamics and network topology without simplifying assumptions, enabling accurate resilience inference and low-dimensional visualization. Now out 👉 #naturecommunications Nature Portfolio https://lnkd.in/eQy2d8nb

28

So glad that this important paper by David Gray Widder is at #FAccT2024! Draws upon 75 interviews with technologists & activists to analyze how value-laden notions of objectivity, quantification, and codification are used to build epistemic power structures in AI ethics practices. These practices legitimize the perspectives and lived experiences of some stakeholders while marginalizing others. Concludes by imploring a turn to "humble technical practices": "I ask those practicing “AI,” and other related quantified sciences widely practiced as value neutral, to help level epistemic power differentials by underscoring the partiality of their work, constructing what I call humble technical practices, which specifically seek to make their epistemic limits clear." Read the full paper here: https://lnkd.in/emxDF2XW #ai #aiethics #techethics #techpolicy #aipolicy #aigovernance

51

6 Comments

#OpenAccess to metadata, such as that relating to citations, authorship and affiliations, is of critical importance to multiple scholarly systems, ecosystems and infrastructures, while assisting with the transparency of publishing processes https://ow.ly/bQIh50SKJBL

3

In this paper, the authors show that by considering the channel state information and #age #of #information (#AoI) of each client, the convergence of FL can further be improved. They formulate a channel-aware joint AoI and diversity-based client scheduling problem as a #constrained #Markov #decision #process (#CMDP). By using Lagrangian index and one-step lookahead approaches, they develop a two-stage online algorithm which is scalable and has a low computational complexity. ---- Manyou Ma, Vincent Wong, Robert Schober More details can be found at this link: https://lnkd.in/eRB8uc_y

3

📚 New Publication Alert 📚 Thrilled to share the news of this new open access publication by Saeema Ahmed-Kristensen and Boyeun Lee titled 'D3 framework: An evidence-based data-driven design framework for new product service development' Highlights: •D3 an evidence-based data-driven design framework for product service development. •Synthesising the use of Data and AI in design-based activities. •Big data is predominantly captured from the in-service phase of a product/service. •Several design activities require both thick and big data. https://lnkd.in/esR4VH4S

1

New paper at #ACM Digital Government: will be useful for web developers making #accessible websites https://lnkd.in/gh2PvN3s

27

𝙲𝚞𝚛𝚛𝚎𝚗𝚝 𝙰𝚙𝚙𝚛𝚘𝚊𝚌𝚑𝚎𝚜 𝚝𝚘 𝙿𝚎𝚎𝚛 𝚁𝚎𝚟𝚒𝚎𝚠 In an effort to align peer review with the general Open Science principles, open peer review (PR) is identified by seven key aspects: open identities, open reports, open participation, open interaction, open pre-review manuscripts, open final-version commenting, and open platforms. 💡 Several innovations to the PR process have been characterized: 1. Tʀᴀᴅɪᴛɪᴏɴᴀʟ ᴠꜱ Tʀᴀɴꜱᴘᴀʀᴇɴᴛ In the most common model, reviewer anonymity allows for impartial decisions, uninfluenced by potential criticism from the authors. In double-anonymised PR, author anonymity limits reviewer bias pertaining to author's gender, country of origin, academic status, or previous publication history.  ✍ The unsigned review reports are made available alongside the published article in transparent PR, where a response by the author may be included. 2. Pʀᴇ ᴠꜱ Pᴏꜱᴛ-ᴘᴜʙʟɪᴄᴀᴛɪᴏɴ Pre-publication PR consists of a formal evaluation of research by editorially invited experts in the relevant field. Not to be confused with pre-PR commenting, which involves the informal commenting or discussion on a publicly available preprint. 👀 Post-publication PR is usually fully public and open, through the intermediation of sites such as Pubpeer—whistleblowing platforms that highlight shortcomings in papers, sometimes leading to retractions and accusations of scientific fraud. 3. Sᴛᴀɢᴇᴅ Results-blind PR is a counter-measure to publication bias, whereupon the results are firtsly made unavailable to the reviewers, making the PR process similar to that of reviewing proposals by scientific grant agencies. 📜 It is an extension of registered reports (which are still restricted mainly to medical fields and psychology), where manuscripts are usually reviewed in two stages—the initial and most important review stage takes place after the study has been designed, but prior to data collection. 4. Cʀᴏᴡᴅꜱᴏᴜʀᴄᴇᴅ Sometimes, the reviewers interact online with the authors and other interested scientists for a more open and collaborative review. Collaborative review involves manuscript assessment wherein referees, editors, and external readers provide interactive comments leading to a consensus decision and a single set of revisions 🔗 Cascading is now widely used, especially by larger publishing houses, and aims to avoid final rejection of a manuscript after peer review by redirecting critically reviewed manuscripts to potentially more suitable journals 5. Dᴇᴄᴏᴜᴘʟᴇᴅ Reports from commercial reviewer platforms can be used to assist in PR. For these portable reviews, the authors pay a company for a standard single-blind review that they can submit with the paper to collaborating journals. 🗂 The post-publication evaluation of significant articles, often through a peer-nominated consortium, can also originate from recommendation services. #scientificwriting #academicpublishing #scholarlyoutput Jigisha Patel Nancy Vesta Peer Review Week

17

3 Comments

Following on from my recent post on #ai and #productivity, which seemed to get a bit of attention... #AI is all about the future, and there's no shortage of pundits willing to make all kinds of predictions about its impact on the future workplace. At times like this, I like to go back to the past. While history isn't always a good predictor of the future, it's better than mere speculation. Introducing a new technology like #ai into an organisation is hard. We know this from decades of experience. My colleague Stefan Hajkowicz recently pointed me to this paper, which says "Information technology projects fail at worrying rates." https://lnkd.in/grsVmMuy We've known this for a long time. My own favourite paper on this topic is by Jonathan Grudin from 1988: Why CSCW Applications Fail. It's a seminal study on why introduction of IT fails. #cscw is a term that's gone out of fashion but it essentially means IT systems that allow people to collaborate on tasks. Through examining four case studies in detail, Grudin concludes that "digital transformation" can fail because: * It can save effort by one group of users but create additional work for others. The typical example is where an IT system is procured by a manager, who gets a benefit, but other workers end up with additional tasks to use this IT system as their needs weren't considered * Existing adoption approaches fail - adoption methods that worked for yesterday's technology may not work for tomorrow's innovations because those introducing the #tech have intuitions based on yesterday not tomorrow * Difficulties in evaluating the #ROI of the tech. "Transformative" technologies are complex and touch many parts of an organisation. This means evaluation is resistant to simple metrics/KPIs and any such won't capture the full picture where the #tech is working and where it isn't After over 35 years, I find these insights are still relevant as organisations try to be part of the #ai #revolution. Do you recognise these factors in your own attempts to adopt #ai? Looking to history is a valuable thing. I know there are some that will say "It's different this time" but if #ai suffers the same socio-technical limitations as technologies that have gone before, shouldn't we learn from that? Full paper: https://lnkd.in/g4-tmQWP

22

1 Comment

The Frontier Model Forum (FMF) released an issue brief that documents early best practices for frontier AI safety evaluations. Read the issue brief here: https://lnkd.in/dqMczrvU Early best practices: (1) Draw on domain expertise (2) Evaluate systems as well as models (3) Consider evaluating marginal risk (4) Account for prompt sensitivity (5) Evaluate against both intended use and adversarial exploitation (6) Take a nuanced approach to evaluation transparency Learn more about model evaluations: T Shevlane and others, ‘Model Evaluation for Extreme Risks’ (arXiv, 2023) <https://lnkd.in/eRb-gRHB> M Phuong and others, ‘Evaluating Frontier Models for Dangerous Capabilities’ (arXiv, 2024) <https://lnkd.in/embgeirH> M Kinniment and others, ‘Evaluating Language-Model Agents on Realistic Autonomous Tasks’ (arXiv, 2023) <https://lnkd.in/emb2HK_Z>

23

2 Comments

Brains and #neuromorphic #systems learn with local learning rules in online-continual learning scenarios. Designing neural networks that learn effectively under these conditions is challenging. Nicholas Alonso and Jeff Krichmar from UC Irvine introduce a #neuralnetwork that implements an effective, principled approach to local, #online-continual #learning on #associative #memory tasks. #Brain-inspired #Computing and #Algorithms Now out 👉 #naturecommunications Nature Portfolio https://lnkd.in/etfqNu2z

62

1 Comment

Good News: The Topic added the journal /Sustainability/, at present, we received more than 10 papers. 😍 #Spatial Decision Support Systems (#SDSSs) are designed around #geospatial data, models, and analytical tools that collectively support human #planning and decision-making procedures in multiple application areas. These areas are constantly evolving to better address existing real-world challenges and find innovative ways forward such as in enabling and facilitating #urban sustainability. https://lnkd.in/gCY7C3BU

2

Excellent article by Alice Evans about the potential great divergence in AI usage Women in one study were 26.4% less likely to use AI and she cites numerous studies that compare within industries and across multiple countries. One interesting thing to note? The biggest discrepancies were when generative AI was banned. Apparently men continued to use it and women's usage dropped quite a bit. How do we get as many people around the table as possible to create #AI4All? #responsibleAI #inclusion

31

8 Comments

I took Meta’s brilliant paper, and asked Anthropic’s Claude 3.5 Sonnet to write a provocative article summarising it from the perspective of what this many mean for closed foundation models. Read below: Although, safety and alignment wise … it’s pretty easy to finetune these things if you know what you’re doing to turn them into subversive comedic monsters 😉🤔😂 Not that I would do such a thing. No sir! (full paper @ https://lnkd.in/d2qs6DPn ) —— Your Move, Sam Altman: Meta's Llama 3.1 Throws Down the Gauntlet In a bold move that has sent shockwaves through the AI community, Meta has unveiled Llama 3.1, a family of LLMs that not only rival the performance of OpenAI's vaunted GPT-4 but do so with a twist: they're giving it away for free. That's right Sam, the game has changed. While OpenAI has been guarding its golden goose behind an API paywall, Meta has decided to release not 1, not 2, but 3 powerhouse models into the wild. We're talking 8B, 70B, and a whopping 405B parameters of open-source AI goodness. It's like they're handing out Lamborghinis at a go-kart race. But it's not just about size (though let's face it, 405 billion parameters is nothing to sneeze at). Llama 3.1 is flexing on multiple fronts: Performance: The 405B model is going toe-to-toe with GPT-4 and Claude 3.5 Sonnet across a smorgasbord of tasks. It's like watching David take on Goliath, except David is also a giant and he's livestreaming the whole thing. Efficiency: While OpenAI's models are busy sipping mai tais in their cushy cloud servers, Llama 3.1’s 8B model can run on a toaster. Okay, maybe not a toaster, but certainly on your average laptop. Democratizing AI? More like starting an AI revolution. Transparency: Meta's not just showing their cards; they're explaining the whole trick. Training process? Here you go. Data curation? Have at it. Evaluation methods? It's all yours. It's so open, it makes OpenAI look like Fort Knox. Safety: They've got Llama Guard watching the gates. It's like they've hired a bouncer for their AI, but instead of keeping people out, it's keeping the AI in check. Multimodal Potential: Vision and speech capabilities are waiting in the wings. It's like Meta's saying, "Oh, you thought language was impressive? Hold my beer." So, what's the catch? There isn't one. Meta's basically dropped a nuke of openness on the AI landscape, and the fallout is going to reshape everything. To Sam Altman and the rest of the closed-source crew, the message is clear: the era of AI gatekeeping is over. Meta's not just inviting everyone to the party; they're giving away the DJ booth, the dance floor, and the whole damn club. In conclusion, Llama 3.1 isn't just a model; it's a manifesto. It's Meta saying, "We're not just in the game; we're changing it." So, Sam, the ball's in your court. Are you going to open up, or are you going to be left behind in a world where state-of-the-art AI is as open as a 24-hour diner? Your move, Sam. Your move.

7

1 Comment

Get #openaccess article at https://lnkd.in/g5muBXFf How to Test and Report #Interaction/#Moderating Effects Here are five key takeaways from our Journal of Management article on reporting interaction effects, which are particularly relevant for #social #science researchers:   1️⃣ Inconsistencies and Gaps in Reporting: Many studies fail to fully describe #interaction #effects, making it difficult to grasp their nature and significance. This inconsistency and lack of detailed reporting can mislead readers about the strength and importance of these effects. 2️⃣ Visualizations Misrepresent Effect Size: Overusing #truncated visual #scales in graphical representations can exaggerate the perceived impact of interaction effects. This practice might lead to undue emphasis on minor findings, steering research and practice in potentially unproductive directions. 3️⃣ Lack of Effect Size Estimates: A significant number of articles do not report a quantifiable measure of interaction effect size, like delta R, which is crucial for assessing the actual impact of the interactions beyond p-values and regression coefficients. 4️⃣ Recommendations for Improvement: We suggest employing full-range visualizations to avoid distortion of effect sizes and recommend including clear, interpretable effect size measures and detailed interpretations of the nature of interaction effects to enhance understanding and application. 5️⃣ Enhancing Theoretical and Practical Relevance: By providing comprehensive details about the strength, nature, and implications of interaction effects, researchers can make their findings more actionable for theory development and practical application, aiding decision-makers in more accurately aligning strategies with evidence-based insights.   These insights can help enhance the research's credibility, rigor, and impact in #management and many other social sciences. Academy of International Business (AIB) Academy of International Business UK & Ireland Chapter Academy of International Business (AIB) Teaching & Education SIG HR Division - Academy of Management Administrative Sciences Association of Canada Africa Academy of Management ANPAD ACEDE (Spanish Academy of Management) British Academy of Management Eastern Academy of Management EUROPEAN ACADEMY OF MANAGEMENT EGOS (European Group for Organizational Studies) @GWSB_alumni Iberoamerican Academy of Management International Council for Small Business (ICSB) Indian Academy of Management - INDAM International Association for Chinese Management Research Management Faculty of Color Association (MFCA) MIDWEST ACADEMY OF MANAGEMENT INC Società Italiana di Management The George Washington University Center for Real Estate and Urban Analysis - The George Washington University School of Business The PhD Project The Strategic Management Society Western Academy of Management (Official Site) Yönetim Akademisi Derneği - Turkish Academy of Management

68

🏛️ The #CIDOC Conceptual Reference Model (CIDOC-CRM) is an ISO Standard ontology for the cultural domain that is used for enabling semantic interoperability between museums, libraries, archives and other cultural institutions. 📄 In this paper, authors analyze the main processes and tasks, and survey recent works that apply #machine_learning techniques for reducing the costs related to CIDOC-CRM-based compliance and interoperability. https://lnkd.in/gqV7BV8X 💡 Moreover, authors introduce their vision for the given topic, and provide a list of open CIDOC-CRM datasets that can be potentially used for machine learning tasks.

4