Training Long-Context Large Language Models (LLMs) is challenging, as hybrid training with long-context and short-context data often leads to workload imbalances. Existing works mainly use data packing to alleviate this issue, but fail to consider imbalanced attention computation and wasted communication overhead. This paper proposes Hierarchical Balance Packing (HBP), which designs a novel batch-construction method and training recipe to address those inefficiencies.

We present two new approaches for point prediction with streaming data. One is based on the Count-Min sketch (CMS) and the other is based on Gaussian process priors with a random bias. These methods are intended for the most general predictive problems where no true model can be usefully formulated for the data stream. In statistical contexts, this is often called the $\mathcal{M}$-open problem class. Under the assumption that the data consists of i.i.d samples from a fixed distribution function $F$, we show that the CMS-based estimates of the distribution function are consistent. We compare our new methods with two established predictors in terms of cumulative $L^1$ error. One is based on the Shtarkov solution (often called the normalized maximum likelihood) in the normal experts setting and the other is based on Dirichlet process priors. These comparisons are for two cases. The first is one-pass meaning that the updating of the predictors is done using the fact that the CMS is a sketch. For predictors that are not one-pass, we use streaming $K$-means to give a representative subset of fixed size that can be updated as data accumulate. Preliminary computational work suggests that the one-pass median version of the CMS method is rarely outperformed by the other methods for sufficiently complex data. We also find that predictors based on Gaussian process priors with random biases perform well. The Shtarkov predictors we use here did not perform as well probably because we were only using the simplest example. The other predictors seemed to perform well mainly when the data did not look like they came from an M-open data generator.

Over the past decade, there have been increasing proclama5ons from diverse stakeholders that humanity is at an inflec5on point due to advances in Ar5ficial Intelligence (AI) technologies (e.g., Crawford, 2017). The general public are condi5oned by this messaging to expect both big (though so far largely non-descript) changes to our lives, including to the way that we learn and teach. Warnings have been also ar5culated regarding whether and how AI might fundamentally change the way we perceive reality, how we form our beliefs, or interact with one another (Bostrom, 2017). More recently, ques5ons started to emerge about AI's transforma5ve poten5al (for beLer or worse) for our func5oning at neurocogni5ve, socio-emo5onal, individual and collec5ve levels (UNESCO, 2022; Pedro, et al., 2019, Porayska-Pomsta, 2023), along with concerns regarding the ethical implica5ons of using AI for suppor5ng human decision-making in contexts that are both high-stakes (e.g., for medical diagnoses or for student assessment) and rela5vely low-stakes, e.g., selec5ng movies on streaming sites. Such hope-fear rhetoric is also present in the context of AI applica5ons to suppor5ng human learning in formal and informal contexts. Recent hopes for AI in educa5on (AIED) largely relate to delivering learning at scale across different geographical and cultural contexts, especially in light of growing global teacher shortages and diminishing funding for educa5on in many countries (UNESCO, 2023). These hopes are increasingly used to fuel poli5cally and market mo5vated discourse about the need to'release teachers from tedious tasks' such as standardised assessments to allow them to focus on the'things that maLer' (Gen5le et al., 2023), or to jus5fy the narrowing of the formal educa5on curricula mainly to STEM subjects.

Driving Automation Systems (DAS) are subject to complex road environments and vehicle behaviors and increasingly rely on sophisticated sensors and Artificial Intelligence (AI). These properties give rise to unique safety faults stemming from specification insufficiencies and technological performance limitations, where sensors and AI introduce errors that vary in magnitude and temporal patterns, posing potential safety risks. The Safety of the Intended Functionality (SOTIF) standard emerges as a promising framework for addressing these concerns, focusing on scenario-based analysis to identify hazardous behaviors and their causes. Although the current standard provides a basic cause-and-effect model and high-level process guidance, it lacks concepts required to identify and evaluate hazardous errors, especially within the context of AI. This paper introduces two key contributions to bridge this gap. First, it defines the SOTIF Temporal Error and Failure Model (STEAM) as a refinement of the SOTIF cause-and-effect model, offering a comprehensive system-design perspective. STEAM refines error definitions, introduces error sequences, and classifies them as error sequence patterns, providing particular relevance to systems employing advanced sensors and AI. Second, this paper proposes the Model-based SOTIF Analysis of Failures and Errors (MoSAFE) method, which allows instantiating STEAM based on system-design models by deriving hazardous error sequence patterns at module level from hazardous behaviors at vehicle level via weakest precondition reasoning. Finally, the paper presents a case study centered on an automated speed-control feature, illustrating the practical applicability of the refined model and the MoSAFE method in addressing complex safety challenges in DAS.

An MRI scan showing the brain of a person with epilepsy.Credit: BSIP/Universal Images Group via Getty Virtual models representing the brains of people with epilepsy could help to enable more-effective treatments of the disease by showing neurosurgeons precisely which zones are responsible for seizures. The models, created using a computational system known as the Virtual Epileptic Patient (VEP), have been developed as part of the Human Brain Project (HBP), a 10-year European initiative focused on digital brain research. The approach is being tested in a clinical trial called EPINOV, to evaluate whether it improves the success rate of epilepsy surgeries. "It's an example of personalized medicine," says Aswin Chari, a neurosurgeon at University College London. VEP uses "the patient's own brain scans [and] the patient's own brainwave-recording data to build a model and improve our understanding of where their seizures are coming from".

The Human Brain Project (HBP) has announced the start of its final phase as an EU-funded FET Flagship. The European Commission has signed a grant agreement to fund the HBP with 150 million Euros from now until 2023. Over the next three years, the project will narrow its focus to advance three core scientific areas – brain networks, their role in consciousness, and artificial neural nets – while expanding its innovative EBRAINS infrastructure. EBRAINS offers the most comprehensive atlas and database on the human brain, directly coupled with powerful computing and simulation tools, to research communities around neuroscience, medicine and technology. Currently transitioning into a sustainable infrastructure, EBRAINS will remain available to the scientific community, as a lasting contribution of the HBP to global scientific progress.

The Helmholtz International BigBrain Analytics and Learning Laboratory (HIBALL) is a collaboration between McGill University and Forschungszentrum Jülich to develop next-generation high-resolution human brain models using cutting-edge Machine- and Deep Learning methods and high-performance computing. HIBALL is based on the high-resolution BigBrain model first published by the Jülich and McGill teams in 2013. Over the next five years, the lab will be funded with a total of up to 6 million Euro by the German Helmholtz Association, Forschungszentrum Jülich, and Healthy Brains, Healthy Lives at McGill University. In 2003, when Jülich neuroscientist Katrin Amunts and her Canadian colleague Alan Evans began scanning 7,404 histological sections of a human brain, it was completely unclear whether it would ever be possible to reconstruct this brain on the computer in three dimensions. At that time, there were no technical possibilities to cope with the huge amount of data.

Dual use and responsible research: ethical challenges' took place at the Karolinska Institute in Stockholm from the 14th to the 17th of November 2018. This workshop showcased the interdisciplinary nature of not only the HBP itself, but the dual use of brain science and the societal impacts this may have. Hence, the purpose of the workshop was to offer a space for discussing both the disciplinary aspects of the HBP, such as neuroscience and medicine, and the wider interdisciplinary aspects such as dual-use and responsible research and innovation (RRI). In order to engage as wide a range of students and researchers as possible with these topics, the workshop was open to all. With lectures covering topics such as the fascinating chemistry behind drug addiction and the revolutionary technology CRISPR that enables geneticists and medical researchers to edit parts of a genome, the interest in the workshop was high.

This workshop aims to provide participants with insights on ethical aspects of dual-use research in neuroscience and Responsible Research and Innovation (RRI). Lectures will be given by some of the world's leading experts on dual-use in neuroscience research, and by active researchers on RRI. The topics covered will include the chemistry of the brain and dual action of drugs, novel incapacitants, ethics awareness and engagement and RRI. An important ingredient of the workshop is the use of team-based learning techniques.

In my previous blog I have explained about linear regression. In today's post I will explain about logistic regression. Consider a scenario where we need to predict a medical condition of a patient (HBP),HAVE HIGH BP or NO HIGH BP, based on some observed symptoms – Age, weight, Issmoking, Systolic value, Diastolic value, RACE, etc.. In this scenario we have to build a model which takes the above mentioned symptoms as input values and HBP as response variable. Note that the response variable (HBP) is a value among a fixed set of classes, HAVE HIGH BP or NO HIGH BP.