Search

One of the most fundamental, and yet relatively less explored, goals in transfer learning is the efficient means of selecting top candidates from a large number of previously trained models (optimized for various "source" tasks) that would perform the best for a new "target" task with a limited amount of data. In this paper, we undertake this goal by developing a novel task-similarity metric (BeST) and an associated method that consistently performs well in identifying the most transferrable source(s) for a given task. In particular, our design employs an innovative quantization-level optimization procedure in the context of classification tasks that yields a measure of similarity between a source model and the given target data. The procedure uses a concept similar to early stopping (usually implemented to train deep neural networks (DNNs) to ensure generalization) to derive a function that approximates the transfer learning mapping without training. The advantage of our metric is that it can be quickly computed to identify the top candidate(s) for a given target task before a computationally intensive transfer operation (typically using DNNs) can be implemented between the selected source and the target task. As such, our metric can provide significant computational savings for transfer learning from a selection of a large number of possible source models. Through extensive experimental evaluations, we establish that our metric performs well over different datasets and varying numbers of data samples.

Robust medical Machine Learning (ML) models have the potential to revolutionize healthcare by accelerating clinical research, improving workflows and outcomes, and producing novel insights or capabilities. Developing such ML models from scratch is cost prohibitive and requires substantial compute, data, and time (e.g., expert labeling). To address these challenges, we introduce Health AI Developer Foundations (HAI-DEF), a suite of pre-trained, domain-specific foundation models, tools, and recipes to accelerate building ML for health applications. The models cover various modalities and domains, including radiology (X-rays and computed tomography), histopathology, dermatological imaging, and audio. These models provide domain specific embeddings that facilitate AI development with less labeled data, shorter training times, and reduced computational costs compared to traditional approaches. In addition, we utilize a common interface and style across these models, and prioritize usability to enable developers to integrate HAI-DEF efficiently. We present model evaluations across various tasks and conclude with a discussion of their application and evaluation, covering the importance of ensuring efficacy, fairness, and equity. Finally, while HAI-DEF and specifically the foundation models lower the barrier to entry for ML in healthcare, we emphasize the importance of validation with problem- and population-specific data for each desired usage setting. This technical report will be updated over time as more modalities and features are added., Comment: 16 pages, 8 figures

Large language models (LLMs) have demonstrated significant success in various domain-specific tasks, with their performance often improving substantially after fine-tuning. However, fine-tuning with real-world data introduces privacy risks. To mitigate these risks, developers increasingly rely on synthetic data generation as an alternative to using real data, as data generated by traditional models is believed to be different from real-world data. However, with the advanced capabilities of LLMs, the distinction between real data and data generated by these models has become nearly indistinguishable. This convergence introduces similar privacy risks for generated data to those associated with real data. Our study investigates whether fine-tuning with LLM-generated data truly enhances privacy or introduces additional privacy risks by examining the structural characteristics of data generated by LLMs, focusing on two primary fine-tuning approaches: supervised fine-tuning (SFT) with unstructured (plain-text) generated data and self-instruct tuning. In the scenario of SFT, the data is put into a particular instruction tuning format used by previous studies. We use Personal Information Identifier (PII) leakage and Membership Inference Attacks (MIAs) on the Pythia Model Suite and Open Pre-trained Transformer (OPT) to measure privacy risks. Notably, after fine-tuning with unstructured generated data, the rate of successful PII extractions for Pythia increased by over 20%, highlighting the potential privacy implications of such approaches. Furthermore, the ROC-AUC score of MIAs for Pythia-6.9b, the second biggest model of the suite, increases over 40% after self-instruct tuning. Our results indicate the potential privacy risks associated with fine-tuning LLMs using generated data, underscoring the need for careful consideration of privacy safeguards in such approaches., Comment: Accepted at 34th USENIX Security Symposium, 2025

It is very important to measure the plasma performances of fusion reactors such as fuel ion density, plasma temperature, fusion power, fast neutron energy, and neutron flux. Various diagnostic systems are needed to measure fast neutron properties in fusion plasmas. Generally, various types of neutron cameras and monitors are used. These systems are sensitive to neutron as well as gamma-ray radiation. Therefore, monitor systems that distinguish between neutron and gamma rays should be used. In this study, the NE-213 liquid scintillation detector was selected, and Pulse Shape Discriminator (PSD) NIM unit was used for the neutron/gamma-ray discrimination. True spectrum was found by unfolding the spectrum obtained by the reaction of neutrons to the protons in the detector with the FERD-PC computer code. In this study, the energy spectrum of fast neutrons produced in a neutron generator was measured by the unfolding method., Comment: 17 pages, 10 figures, 2 tables

This study proposes a simple architecture for Enterprise application for Large Language Models (LLMs) for role based security and NATO clearance levels. Our proposal aims to address the limitations of current LLMs in handling security and information access. The proposed architecture could be used while utilizing Retrieval-Augmented Generation (RAG) and fine tuning of Mixture of experts models (MoE). It could be used only with RAG, or only with MoE or with both of them. Using roles and security clearance level of the user, documents in RAG and experts in MoE are filtered. This way information leakage is prevented.

Current Motor Imagery Brain-Computer Interfaces (MI-BCI) require a lengthy and monotonous training procedure to train both the system and the user. Considering many users struggle with effective control of MI-BCI systems, a more user-centered approach to training might help motivate users and facilitate learning, alleviating inefficiency of the BCI system. With the increase of BCI-controlled games, researchers have suggested using game principles for BCI training, as games are naturally centered on the player. This review identifies and evaluates the application of game design elements to MI-BCI training, a process known as gamification. Through a systematic literature search, we examined how MI-BCI training protocols have been gamified and how specific game elements impacted the training outcomes. We identified 86 studies that employed gamified MI-BCI protocols in the past decade. The prevalence and reported effects of individual game elements on user experience and performance were extracted and synthesized. Results reveal that MI-BCI training protocols are most often gamified by having users move an avatar in a virtual environment that provides visual feedback. Furthermore, in these virtual environments, users were provided with goals that guided their actions. Using gamification, the reviewed protocols allowed users to reach effective MI-BCI control, with studies reporting positive effects of four individual elements on user performance and experience, namely: feedback, avatars, assistance, and social interaction. Based on these elements, this review makes current and future recommendations for effective gamification, such as the use of virtual reality and adaptation of game difficulty to user skill level.

With the development of edge networks and mobile computing, the need to serve heterogeneous data sources at the network edge requires the design of new distributed machine learning mechanisms. As a prevalent approach, Federated Learning (FL) employs parameter-sharing and gradient-averaging between clients and a server. Despite its many favorable qualities, such as convergence and data-privacy guarantees, it is well-known that classic FL fails to address the challenge of data heterogeneity and computation heterogeneity across clients. Most existing works that aim to accommodate such sources of heterogeneity stay within the FL operation paradigm, with modifications to overcome the negative effect of heterogeneous data. In this work, as an alternative paradigm, we propose a Multi-Task Split Learning (MTSL) framework, which combines the advantages of Split Learning (SL) with the flexibility of distributed network architectures. In contrast to the FL counterpart, in this paradigm, heterogeneity is not an obstacle to overcome, but a useful property to take advantage of. As such, this work aims to introduce a new architecture and methodology to perform multi-task learning for heterogeneous data sources efficiently, with the hope of encouraging the community to further explore the potential advantages we reveal. To support this promise, we first show through theoretical analysis that MTSL can achieve fast convergence by tuning the learning rate of the server and clients. Then, we compare the performance of MTSL with existing multi-task FL methods numerically on several image classification datasets to show that MTSL has advantages over FL in training speed, communication cost, and robustness to heterogeneous data.

In this paper, we study a natural policy gradient method based on recurrent neural networks (RNNs) for partially-observable Markov decision processes, whereby RNNs are used for policy parameterization and policy evaluation to address curse of dimensionality in non-Markovian reinforcement learning. We present finite-time and finite-width analyses for both the critic (recurrent temporal difference learning), and correspondingly-operated recurrent natural policy gradient method in the near-initialization regime. Our analysis demonstrates the efficiency of RNNs for problems with short-term memory with explicit bounds on the required network widths and sample complexity, and points out the challenges in the case of long-term dependencies.

Many clinical tasks require an understanding of specialized data, such as medical images and genomics, which is not typically found in general-purpose large multimodal models. Building upon Gemini's multimodal models, we develop several models within the new Med-Gemini family that inherit core capabilities of Gemini and are optimized for medical use via fine-tuning with 2D and 3D radiology, histopathology, ophthalmology, dermatology and genomic data. Med-Gemini-2D sets a new standard for AI-based chest X-ray (CXR) report generation based on expert evaluation, exceeding previous best results across two separate datasets by an absolute margin of 1% and 12%, where 57% and 96% of AI reports on normal cases, and 43% and 65% on abnormal cases, are evaluated as "equivalent or better" than the original radiologists' reports. We demonstrate the first ever large multimodal model-based report generation for 3D computed tomography (CT) volumes using Med-Gemini-3D, with 53% of AI reports considered clinically acceptable, although additional research is needed to meet expert radiologist reporting quality. Beyond report generation, Med-Gemini-2D surpasses the previous best performance in CXR visual question answering (VQA) and performs well in CXR classification and radiology VQA, exceeding SoTA or baselines on 17 of 20 tasks. In histopathology, ophthalmology, and dermatology image classification, Med-Gemini-2D surpasses baselines across 18 out of 20 tasks and approaches task-specific model performance. Beyond imaging, Med-Gemini-Polygenic outperforms the standard linear polygenic risk score-based approach for disease risk prediction and generalizes to genetically correlated diseases for which it has never been trained. Although further development and evaluation are necessary in the safety-critical medical domain, our results highlight the potential of Med-Gemini across a wide range of medical tasks.

Navigation and trajectorial estimation of maritime vessels are contingent upon the context of positional accuracy. Even the smallest deviations in the estimation of a given vessel may result in detrimental consequences in terms of economic and ecologic quotients. To ensure an agile and precise environment for maritime vessel positional estimation, preexisting marine radar technologies can be utilized in a way that ensures a higher level of precision compared to GNSS-based identification and positioning. In this paper, we present a positional optimization for radarbased vessel navigation systems that utilize the installment of vessel detection sensors. The main objective of this research is to employ as fewer sensors as possible while preserving the attainable error threshold for positioning that is defined by International Maritime Organization (IMO). Our approach leads most of the time to a positioning error of up to 5 m along shorelines and rivers and up to 50 m along open coastal regions., Comment: 9 pages, 7 figures, 3 tables, WoWMoM 2024

A new digital twin (DT) framework with optimal sensor placement (OSP) is proposed to accurately calculate the modal responses and identify the damage ratios of the offshore jacket platforms. The proposed damage identification framework consists of two models (namely one OSP model and one damage identification model). The OSP model adopts the multi-objective Lichtenberg algorithm (MOLA) to perform the sensor number/location optimization to make a good balance between the sensor cost and the modal calculation accuracy. In the damage identification model, the Markov Chain Monte Carlo (MCMC)-Bayesian method is developed to calculate the structural damage ratios based on the modal information obtained from the sensory measurements, where the uncertainties of the structural parameters are quantified. The proposed method is validated using an offshore jacket platform, and the analysis results demonstrate efficient identification of the structural damage location and severity.

We establish homogenization for nondegenerate viscous Hamilton-Jacobi equations in one space dimension when the diffusion coefficient $a(x,\omega) > 0$ and the Hamiltonian $H(p,x,\omega)$ are general stationary ergodic processes in $x$. Our result is valid under mild regularity assumptions on $a$ and $H$ plus standard coercivity and growth assumptions (in $p$) on the latter. In particular, we impose neither any additional condition on the law of the media nor any shape restriction on the graph of $p\mapsto H(p,x,\omega)$. Our approach consists of two main steps: (i) constructing a suitable candidate $\overline{H}$ for the effective Hamiltonian; (ii) proving homogenization. In the first step, we work with the set $E$ of all points at which $\overline{H}$ is naturally determined by correctors with stationary derivatives. We prove that $E$ is a closed subset of $\mathbb{R}$ that is unbounded from above and below, and, if $E\neq\mathbb{R}$, then $\overline{H}$ can be extended continuously to $\mathbb{R}$ by setting it to be constant on each connected component of $E^c$. In the second step, we use a key bridging lemma, comparison arguments and several general results to verify that homogenization holds with this $\overline{H}$ as the effective Hamiltonian., Comment: 20 pages

We introduce SwiftCache, a "fresh" learning-based caching framework designed for content distribution networks (CDNs) featuring distributed front-end local caches and a dynamic back-end database. Users prefer the most recent version of the dynamically updated content, while the local caches lack knowledge of item popularity and refresh rates. We first explore scenarios with requests arriving at a local cache following a Poisson process, whereby we prove that the optimal policy features a threshold-based structure with updates occurring solely at request arrivals. Leveraging these findings, SwiftCache is proposed as a model-based learning framework for dynamic content caching. The simulation demonstrates near-optimal cost for Poisson process arrivals and strong performance with limited cache sizes. For more general environments, we present a model-free Reinforcement Learning (RL) based caching policy without prior statistical assumptions. The model-based policy performs well compared to the model-free policy when the variance of interarrival times remains moderate. However, as the variance increases, RL slightly outperforms model-based learning at the cost of longer training times, and higher computational resource consumption. Model-based learning's adaptability to environmental changes without retraining positions it as a practical choice for dynamic network environments. Distributed edge caches can utilize this approach in a decentralized manner to effectively meet the evolving behaviors of users., Comment: arXiv admin note: text overlap with arXiv:2401.03613

We analyze recurrent neural networks with diagonal hidden-to-hidden weight matrices, trained with gradient descent in the supervised learning setting, and prove that gradient descent can achieve optimality \emph{without} massive overparameterization. Our in-depth nonasymptotic analysis (i) provides improved bounds on the network size $m$ in terms of the sequence length $T$, sample size $n$ and ambient dimension $d$, and (ii) identifies the significant impact of long-term dependencies in the dynamical system on the convergence and network width bounds characterized by a cutoff point that depends on the Lipschitz continuity of the activation function. Remarkably, this analysis reveals that an appropriately-initialized recurrent neural network trained with $n$ samples can achieve optimality with a network size $m$ that scales only logarithmically with $n$. This sharply contrasts with the prior works that require high-order polynomial dependency of $m$ on $n$ to establish strong regularity conditions. Our results are based on an explicit characterization of the class of dynamical systems that can be approximated and learned by recurrent neural networks via norm-constrained transportation mappings, and establishing local smoothness properties of the hidden state with respect to the learnable parameters.

Navigation has been classically solved in robotics through the combination of SLAM and planning. More recently, beyond waypoint planning, problems involving significant components of (visual) high-level reasoning have been explored in simulated environments, mostly addressed with large-scale machine learning, in particular RL, offline-RL or imitation learning. These methods require the agent to learn various skills like local planning, mapping objects and querying the learned spatial representations. In contrast to simpler tasks like waypoint planning (PointGoal), for these more complex tasks the current state-of-the-art models have been thoroughly evaluated in simulation but, to our best knowledge, not yet in real environments. In this work we focus on sim2real transfer. We target the challenging Multi-Object Navigation (Multi-ON) task and port it to a physical environment containing real replicas of the originally virtual Multi-ON objects. We introduce a hybrid navigation method, which decomposes the problem into two different skills: (1) waypoint navigation is addressed with classical SLAM combined with a symbolic planner, whereas (2) exploration, semantic mapping and goal retrieval are dealt with deep neural networks trained with a combination of supervised learning and RL. We show the advantages of this approach compared to end-to-end methods both in simulation and a real environment and outperform the SOTA for this task.

This paper proposes an elementary solution to a special case of finding all perfect squares that can be written as sum of consecutive integer cubes. It is shown that there are no non-trivial solutions if the perfect square is a prime power, i.e., it is divisible by two different primes if a non-trivial one exists. Solution mostly depends on $v_{p}(x)$ and general forms of Pythagorean triples.

We introduce a framework and optimal `fresh' caching for a content distribution network (CDN) comprising a front-end local cache and a back-end database. The data content is dynamically updated at a back-end database and end-users are interested in the most-recent version of that content. We formulate the average cost minimization problem that captures the system's cost due to the service of aging content as well as the regular cache update cost. We consider the cost minimization problem from two individual perspectives based on the available information to either side of the CDN: the back-end database perspective and the front-end local cache perspective. For the back-end database, the instantaneous version of content is observable but the exact demand is not. Caching decisions made by the back-end database are termed `push-based caching'. For the front-end local cache, the age of content version in the cache is not observable, yet the instantaneous demand is. Caching decisions made by the front-end local cache are termed `pull-based caching'. Our investigations reveal which type of information, updates, or demand dynamic, is of higher value towards achieving the minimum cost based on other network parameters including content popularity, update rate, and demand intensity., Comment: 14 pages, 7 figures