Washington State School Directors' Association ’s Post

🌟 Excited to share this new blog post on Data-Free Knowledge Distillation (DFKD) and our proposed method, Teacher-Agnostic Data-Free Knowledge Distillation (TA-DFKD). Our team's research explores the challenges of existing DFKD methods and introduces a more robust approach that ensures stable performance across various teacher models. Check out the full post here: https://bit.ly/48kCNP7 #knowledgeDistillation #MachineLearning #DataScience

A handy guide to using research evidence https://lnkd.in/gC8JDSAi

With the new school year around the corner, State Education Agencies can benefit from this NCEO tool, which provides answers to frequently asked questions about AA-AAAS. For example: Must the state publicly post the actual LEA justification (with personal information redacted or deleted)? Learn more here: https://z.umn.edu/AATK5 

Exciting indicator that federal policy may be shifting from the outdated "research-practice pipeline" to a more impactful R&D focus. The bipartisan New Essential Education Discoveries (NEED) Act has been introduced in the Senate by Senator Michael Bennet (D-CO) and Senator John Cornyn (R-TX). If passed, the NEED Act would authorize a National Center for Advanced Development in Education (NCADE) to advance informed-risk high-reward, education R&D projects. It would support interdisciplinary teams to solve big, complex challenges in education through new research, rapid testing, and the iterative development of new tools and approaches. The NEED Act would also strengthen and modernize Statewide Longitudinal Data Systems, with good data as the backbone of good R&D. #edresearch  #eddata #education In Tandem

One of our clients is doing a really interesting research project entitled "How can the functions of a Multi Academy Trust in England be measured for efficacy?" and is looking for input from some MAT CFOs, please see further information below. The purpose of this research is to explore something of a void in the realm of MATs in regards to modelling and benchmarking financial efficacy. It has taken around 2 decades of academisation before we have seen clarity on what a ‘high quality trust’ looks like and for MATs to be considered under the Ofsted accountability framework. However, there is still little research around the most efficient models for funding central services and how trusts can or should model this in a meaningful way to demonstrate value for money. I will be using an array of secondary data that exists in the public domain but am also seeking more detailed primary data through surveys completed by MAT leaders to contextualise their approach to managing central services. To participate, please complete the forms using the 2 links below: The consent form which reassures of anonymity in data collated: https://lnkd.in/ebnSmN9W The questionnaire which should take around 20 minutes: https://lnkd.in/eze_4epm

What’s the Department of Education got planned for this coming year? Let’s take a look at the tentative regulatory timeline, keyword *tentative* of course. https://lnkd.in/eFpn-jQ3

📉 Mitigating Nonproductive Time: A Novel Algorithm for Dsl Fault Detection I'm delighted to announce the publication of our scientific paper, a result of my work at SLB. I would like to thank all the authors and reviewers, especially my day-to-day coworker, Shruthi Shetty, and my manager, Abhinav Kohar, for all the unwavering support provided so far. It's my first paper published in the industry. It results from the first end-to-end project I've worked on this past year. A big shoutout to SLB for this opportunity 🎉 🔗 Link to the paper: https://lnkd.in/dXbS22df #datascience #statistics #algorithms

We wish research evidence always gave definitive answers. It would make our job as a Research School a lot easier, and make it easier for schools to implement the best approaches. But more often than not, when asked a question about evidence, the answer is ‘it’s complicated’. Instead of seeking certainty, we can ask:  Under which circumstances was this effective?  Under which circumstances was this ineffective?  What can we say for sure?  What don’t we know? And being able to reflect on these questions is more beneficial to a nuanced understanding of the evidence than bolder claims. The questions signpost us towards reading further around the subject. We can weigh up evidence, and use that to help us decide what would be the most fruitful thing to focus on. We place our best bets’. It’s important that when we communicate these  ‘best bets’, that we acknowledge the basis for our claims, and the ways we might be wrong. Daniel Willingham, in When Can You Trust the Experts, sums it up:  “Indeed, when certainty is not available, a plausible guess is the best we’re going to do.” 

The "User-friendly Introduction to PAC-Bayes Bounds" by my friend Pierre Alquier is now officially published! I strongly recommend it if you want to know what PAC-Bayes is about. You can purchase it or download a free version here (you can also find it on arxiv): By Pierre Alquier https://lnkd.in/edYfqy6X

Summer is heating up on #OA policy: June 18, the RFI on the NIH draft public access policy was posted. On the 24th, the full report (sort of) that Congress asked for was posted. TLDR? .... Are reasonable publication fees to be allowed? *Yes.* (Unless 55 days of comments sways this). What's reasonable? Guidelines offered. Thanks to Lisa Janicke Hinchliffe and Ashley Farley who shared their first look on X and pointed me to the RFI...been waiting (link in comments) Access to publicly funded research has value, but does the NIH know what they are spending? (can you guess?) Read Kent Anderson's take (subscription required) and see the full report (link in comments). Also this week, Mike Taylor posted a preprint on research into the OA Citation Advantage and OA Altmetric advantage (link in comments) - the answer: Yes there are advantages, but it's complicated. and last week in the UK regarding REF and removing the burden of OA...(more links below)....