Poonam Yadav

Kavan Fatehi, Mostafa Rahmani Ghourtani, Amir Sonee et al.

Sixth-generation (6G) radio access networks (RANs) must enforce strict service-level agreements (SLAs) for heterogeneous slices, yet sudden latency spikes remain difficult to diagnose and resolve with conventional deep reinforcement learning (DRL) or explainable RL (XRL). We propose \emph{Attention-Enhanced Multi-Agent Proximal Policy Optimization (AE-MAPPO)}, which integrates six specialized attention mechanisms into multi-agent slice control and surfaces them as zero-cost, faithful explanations. The framework operates across O-RAN timescales with a three-phase strategy: predictive, reactive, and inter-slice optimization. A URLLC case study shows AE-MAPPO resolves a latency spike in $18$ms, restores latency to $0.98$ms with $99.9999\%$ reliability, and reduces troubleshooting time by $93\%$ while maintaining eMBB and mMTC continuity. These results confirm AE-MAPPO's ability to combine SLA compliance with inherent interpretability, enabling trustworthy and real-time automation for 6G RAN slicing.

Bertie Vidgen, Adarsh Agrawal, Ahmed M. Ahmed et al. · deepmind, oxford

This paper introduces v0.5 of the AI Safety Benchmark, which has been created by the MLCommons AI Safety Working Group. The AI Safety Benchmark has been designed to assess the safety risks of AI systems that use chat-tuned language models. We introduce a principled approach to specifying and constructing the benchmark, which for v0.5 covers only a single use case (an adult chatting to a general-purpose assistant in English), and a limited set of personas (i.e., typical users, malicious users, and vulnerable users). We created a new taxonomy of 13 hazard categories, of which 7 have tests in the v0.5 benchmark. We plan to release version 1.0 of the AI Safety Benchmark by the end of 2024. The v1.0 benchmark will provide meaningful insights into the safety of AI systems. However, the v0.5 benchmark should not be used to assess the safety of AI systems. We have sought to fully document the limitations, flaws, and challenges of v0.5. This release of v0.5 of the AI Safety Benchmark includes (1) a principled approach to specifying and constructing the benchmark, which comprises use cases, types of systems under test (SUTs), language and context, personas, tests, and test items; (2) a taxonomy of 13 hazard categories with definitions and subcategories; (3) tests for seven of the hazard categories, each comprising a unique set of test items, i.e., prompts. There are 43,090 test items in total, which we created with templates; (4) a grading system for AI systems against the benchmark; (5) an openly available platform, and downloadable tool, called ModelBench that can be used to evaluate the safety of AI systems on the benchmark; (6) an example evaluation report which benchmarks the performance of over a dozen openly available chat-tuned language models; (7) a test specification for the benchmark.

Jinkun Zhang, Poonam Yadav

Partitioned DNN inference is a promising approach for latency-sensitive intelligent services in edge networks, since it allows different parts of a model to be executed across end devices, edge servers, and the cloud. However, in a multi-hop edge network, partition placement and inference traffic routing are inherently coupled: raw inputs, intermediate features, and final outputs may have very different sizes, while candidate nodes also differ in computation capability. In addition, both communication and computation delays can become congestion-dependent under load. In this paper, we study joint partition placement and routing for fixed-partition DNN inference over heterogeneous multi-hop edge networks. We consider a small number of DNN partitions, each placed at exactly one node without replication, and formulate a congestion-aware mixed discrete--continuous optimization problem that captures both routing and execution costs. To solve it, we develop a practical alternating framework that couples partition placement with congestion-aware forwarding updates. Through numerical evaluation on hierarchical, regular, synthetic irregular, and real backbone-inspired topologies, we show that split flexibility is particularly important in IoT--edge--cloud settings, while congestion-aware refinement becomes increasingly beneficial as the offered load grows. We further illustrate how the preferred operating point depends on the communication--computation tradeoff.

Alexandros Karargyris, Renato Umeton, Micah J. Sheller et al.

Medical AI has tremendous potential to advance healthcare by supporting the evidence-based practice of medicine, personalizing patient treatment, reducing costs, and improving provider and patient experience. We argue that unlocking this potential requires a systematic way to measure the performance of medical AI models on large-scale heterogeneous data. To meet this need, we are building MedPerf, an open framework for benchmarking machine learning in the medical domain. MedPerf will enable federated evaluation in which models are securely distributed to different facilities for evaluation, thereby empowering healthcare organizations to assess and verify the performance of AI models in an efficient and human-supervised process, while prioritizing privacy. We describe the current challenges healthcare and AI communities face, the need for an open platform, the design philosophy of MedPerf, its current implementation status, and our roadmap. We call for researchers and organizations to join us in creating the MedPerf open benchmarking platform.

Roman Kolcun, Diana Andreea Popescu, Vadim Safronov et al.

Internet-of-Things (IoT) devices are known to be the source of many security problems, and as such, they would greatly benefit from automated management. This requires robustly identifying devices so that appropriate network security policies can be applied. We address this challenge by exploring how to accurately identify IoT devices based on their network behavior, while leveraging approaches previously proposed by other researchers. We compare the accuracy of four different previously proposed machine learning models (tree-based and neural network-based) for identifying IoT devices. We use packet trace data collected over a period of six months from a large IoT test-bed. We show that, while all models achieve high accuracy when evaluated on the same dataset as they were trained on, their accuracy degrades over time, when evaluated on data collected outside the training set. We show that on average the models' accuracy degrades after a couple of weeks by up to 40 percentage points (on average between 12 and 21 percentage points). We argue that, in order to keep the models' accuracy at a high level, these need to be continuously updated.

Roman Kolcun, Diana Andreea Popescu, Vadim Safronov et al.

Internet-of-Things (IoT) devices are known to be the source of many security problems, and as such they would greatly benefit from automated management. This requires robustly identifying devices so that appropriate network security policies can be applied. We address this challenge by exploring how to accurately identify IoT devices based on their network behavior, using resources available at the edge of the network. In this paper, we compare the accuracy of five different machine learning models (tree-based and neural network-based) for identifying IoT devices by using packet trace data from a large IoT test-bed, showing that all models need to be updated over time to avoid significant degradation in accuracy. In order to effectively update the models, we find that it is necessary to use data gathered from the deployment environment, e.g., the household. We therefore evaluate our approach using hardware resources and data sources representative of those that would be available at the edge of the network, such as in an IoT deployment. We show that updating neural network-based models at the edge is feasible, as they require low computational and memory resources and their structure is amenable to being updated. Our results show that it is possible to achieve device identification and categorization with over 80% and 90% accuracy respectively at the edge.

Poonam Yadav, Angelo Feraudo, Budi Arief et al.

The popularity of the Internet of Things (IoT) devices makes it increasingly important to be able to fingerprint them, for example in order to detect if there are misbehaving or even malicious IoT devices in one's network. The aim of this paper is to provide a systematic categorisation of machine learning augmented techniques that can be used for fingerprinting IoT devices. This can serve as a baseline for comparing various IoT fingerprinting mechanisms, so that network administrators can choose one or more mechanisms that are appropriate for monitoring and maintaining their network. We carried out an extensive literature review of existing papers on fingerprinting IoT devices -- paying close attention to those with machine learning features. This is followed by an extraction of important and comparable features among the mechanisms outlined in those papers. As a result, we came up with a key set of terminologies that are relevant both in the fingerprinting context and in the IoT domain. This enabled us to construct a framework called IDWork, which can be used for categorising existing IoT fingerprinting mechanisms in a way that will facilitate a coherent and fair comparison of these mechanisms. We found that the majority of the IoT fingerprinting mechanisms take a passive approach -- mainly through network sniffing -- instead of being intrusive and interactive with the device of interest. Additionally, a significant number of the surveyed mechanisms employ both static and dynamic approaches, in order to benefit from complementary features that can be more robust against certain attacks such as spoofing and replay attacks.

Colby R. Banbury, Vijay Janapa Reddi, Max Lam et al.

Recent advancements in ultra-low-power machine learning (TinyML) hardware promises to unlock an entirely new class of smart applications. However, continued progress is limited by the lack of a widely accepted benchmark for these systems. Benchmarking allows us to measure and thereby systematically compare, evaluate, and improve the performance of systems and is therefore fundamental to a field reaching maturity. In this position paper, we present the current landscape of TinyML and discuss the challenges and direction towards developing a fair and useful hardware benchmark for TinyML workloads. Furthermore, we present our four benchmarks and discuss our selection methodology. Our viewpoints reflect the collective thoughts of the TinyMLPerf working group that is comprised of over 30 organizations.

Poonam Yadav, John Darlington

In recent years, citizen science has grown in popularity due to a number of reasons, including the emphasis on informal learning and creativity potential associated with these initiatives. Citizen science projects address research questions from various domains, ranging from Ecology to Astronomy. Due to the advancement of communication technologies, which makes outreach and engagement of wider communities easier, scientists are keen to turn their own research into citizen science projects. However, the development, deployment and management of these projects remains challenging. One of the most important challenges is building the project itself. There is no single tool or framework, which guides the step-by-step development of the project, since every project has specific characteristics, such as geographical constraints or volunteers' mode of participation. Therefore, in this article, we present a series of conceptual frameworks for categorisation, decision and deployment, which guide a citizen science project creator in every step of creating a new project starting from the research question to project deployment. The frameworks are designed with consideration to the properties of already existing citizen science projects and could be easily extended to include other dimensions, which are not currently perceived.

Poonam Yadav, John Darlington

Online Citizen Science platforms are good examples of socio-technical systems where technology-enabled interactions occur between scientists and the general public (volunteers). Citizen Science platforms usually host multiple Citizen Science projects, and allow volunteers to choose the ones to participate in. Recent work in the area has demonstrated a positive feedback loop between participation and learning and creativity in Citizen Science projects, which is one of the motivating factors both for scientists and the volunteers. This emphasises the importance of creating successful Citizen Science platforms, which support this feedback process, and enable enhanced learning and creativity to occur through knowledge sharing and diverse participation. In this paper, we discuss how scientists' and volunteers' motivation and participation influence the design of Citizen Science platforms. We present our summary as guidelines for designing these platforms as user-inspired socio-technical systems. We also present the case-studies on popular Citizen Science platforms, including our own CitizenGrid platform, developed as part of the CCL EU project, as well as Zooniverse, World Community Grid, CrowdCrafting and EpiCollect+ to see how closely these platforms follow our proposed guidelines and how these may be further improved to incorporate the creativity enabled by the collective knowledge sharing.

Poonam Yadav

Automated face recognition and identification softwares are becoming part of our daily life; it finds its abode not only with Facebook's auto photo tagging, Apple's iPhoto, Google's Picasa, Microsoft's Kinect, but also in Homeland Security Department's dedicated biometric face detection systems. Most of these automatic face identification systems fail where the effects of aging come into the picture. Little work exists in the literature on the subject of face prediction that accounts for aging, which is a vital part of the computer face recognition systems. In recent years, individual face components' (e.g. eyes, nose, mouth) features based matching algorithms have emerged, but these approaches are still not efficient. Therefore, in this work we describe a Face Prediction Model (FPM), which predicts human face aging or growth related image variation using Principle Component Analysis (PCA) and Artificial Neural Network (ANN) learning techniques. The FPM captures the facial changes, which occur with human aging and predicts the facial image with a few years of gap with an acceptable accuracy of face matching from 76 to 86%.