Mete Özay

Junjie Hu, Chenyou Fan, Mete Ozay et al.

We study a practical yet hasn't been explored problem: how a drone can perceive in an environment from different flight heights. Unlike autonomous driving, where the perception is always conducted from a ground viewpoint, a flying drone may flexibly change its flight height due to specific tasks, requiring the capability for viewpoint invariant perception. Tackling the such problem with supervised learning will incur tremendous costs for data annotation of different flying heights. On the other hand, current semi-supervised learning methods are not effective under viewpoint differences. In this paper, we introduce the ground-to-aerial perception knowledge transfer and propose a progressive semi-supervised learning framework that enables drone perception using only labeled data of ground viewpoint and unlabeled data of flying viewpoints. Our framework has four core components: i) a dense viewpoint sampling strategy that splits the range of vertical flight height into a set of small pieces with evenly-distributed intervals, ii) nearest neighbor pseudo-labeling that infers labels of the nearest neighbor viewpoint with a model learned on the preceding viewpoint, iii) MixView that generates augmented images among different viewpoints to alleviate viewpoint differences, and iv) a progressive distillation strategy to gradually learn until reaching the maximum flying height. We collect a synthesized and a real-world dataset, and we perform extensive experimental analyses to show that our method yields 22.2% and 16.9% accuracy improvement for the synthesized dataset and the real world. Code and datasets are available on https://github.com/FreeformRobotics/Progressive-Self-Distillation-for-Ground-to-Aerial-Perception-Knowledge-Transfer.

Junjie Hu, Chenyu Bao, Mete Ozay et al.

Depth completion aims at predicting dense pixel-wise depth from an extremely sparse map captured from a depth sensor, e.g., LiDARs. It plays an essential role in various applications such as autonomous driving, 3D reconstruction, augmented reality, and robot navigation. Recent successes on the task have been demonstrated and dominated by deep learning based solutions. In this article, for the first time, we provide a comprehensive literature review that helps readers better grasp the research trends and clearly understand the current advances. We investigate the related studies from the design aspects of network architectures, loss functions, benchmark datasets, and learning strategies with a proposal of a novel taxonomy that categorizes existing methods. Besides, we present a quantitative comparison of model performance on three widely used benchmarks, including indoor and outdoor datasets. Finally, we discuss the challenges of prior works and provide readers with some insights for future research directions.

Hayder Elesedy, Pedro M. Esperança, Silviu Vlad Oprea et al. · amazon-science

Guardrails have emerged as an alternative to safety alignment for content moderation of large language models (LLMs). Existing model-based guardrails have not been designed for resource-constrained computational portable devices, such as mobile phones, more and more of which are running LLM-based applications locally. We introduce LoRA-Guard, a parameter-efficient guardrail adaptation method that relies on knowledge sharing between LLMs and guardrail models. LoRA-Guard extracts language features from the LLMs and adapts them for the content moderation task using low-rank adapters, while a dual-path design prevents any performance degradation on the generative task. We show that LoRA-Guard outperforms existing approaches with 100-1000x lower parameter overhead while maintaining accuracy, enabling on-device content moderation.

Junjie Hu, Chenyou Fan, Mete Ozay et al.

We study data-free knowledge distillation (KD) for monocular depth estimation (MDE), which learns a lightweight model for real-world depth perception tasks by compressing it from a trained teacher model while lacking training data in the target domain. Owing to the essential difference between image classification and dense regression, previous methods of data-free KD are not applicable to MDE. To strengthen its applicability in real-world tasks, in this paper, we propose to apply KD with out-of-distribution simulated images. The major challenges to be resolved are i) lacking prior information about scene configurations of real-world training data and ii) domain shift between simulated and real-world images. To cope with these difficulties, we propose a tailored framework for depth distillation. The framework generates new training samples for embracing a multitude of possible object arrangements in the target domain and utilizes a transformation network to efficiently adapt them to the feature statistics preserved in the teacher model. Through extensive experiments on various depth estimation models and two different datasets, we show that our method outperforms the baseline KD by a good margin and even achieves slightly better performance with as few as 1/6 of training images, demonstrating a clear superiority.

Umberto Michieli, Pablo Peso Parada, Mete Ozay

Keyword Spotting (KWS) models on embedded devices should adapt fast to new user-defined words without forgetting previous ones. Embedded devices have limited storage and computational resources, thus, they cannot save samples or update large models. We consider the setup of embedded online continual learning (EOCL), where KWS models with frozen backbone are trained to incrementally recognize new words from a non-repeated stream of samples, seen one at a time. To this end, we propose Temporal Aware Pooling (TAP) which constructs an enriched feature space computing high-order moments of speech features extracted by a pre-trained backbone. Our method, TAP-SLDA, updates a Gaussian model for each class on the enriched feature space to effectively use audio representations. In experimental analyses, TAP-SLDA outperforms competitors on several setups, backbones, and baselines, bringing a relative average gain of 11.3% on the GSC dataset.

Umberto Michieli, Mete Ozay

Vision systems mounted on home robots need to interact with unseen classes in changing environments. Robots have limited computational resources, labelled data and storage capability. These requirements pose some unique challenges: models should adapt without forgetting past knowledge in a data- and parameter-efficient way. We characterize the problem as few-shot (FS) online continual learning (OCL), where robotic agents learn from a non-repeated stream of few-shot data updating only a few model parameters. Additionally, such models experience variable conditions at test time, where objects may appear in different poses (e.g., horizontal or vertical) and environments (e.g., day or night). To improve robustness of CL agents, we propose RobOCLe, which; 1) constructs an enriched feature space computing high order statistical moments from the embedded features of samples; and 2) computes similarity between high order statistics of the samples on the enriched feature space, and predicts their class labels. We evaluate robustness of CL models to train/test augmentations in various cases. We show that different moments allow RobOCLe to capture different properties of deformations, providing higher robustness with no decrease of inference speed.

Kirill Paramonov, Jia-Xing Zhong, Umberto Michieli et al.

In this paper, we address a recent trend in robotic home appliances to include vision systems on personal devices, capable of personalizing the appliances on the fly. In particular, we formulate and address an important technical task of personal object search, which involves localization and identification of personal items of interest on images captured by robotic appliances, with each item referenced only by a few annotated images. The task is crucial for robotic home appliances and mobile systems, which need to process personal visual scenes or to operate with particular personal objects (e.g., for grasping or navigation). In practice, personal object search presents two main technical challenges. First, a robot vision system needs to be able to distinguish between many fine-grained classes, in the presence of occlusions and clutter. Second, the strict resource requirements for the on-device system restrict the usage of most state-of-the-art methods for few-shot learning and often prevent on-device adaptation. In this work, we propose Swiss DINO: a simple yet effective framework for one-shot personal object search based on the recent DINOv2 transformer model, which was shown to have strong zero-shot generalization properties. Swiss DINO handles challenging on-device personalized scene understanding requirements and does not require any adaptation training. We show significant improvement (up to 55%) in segmentation and recognition accuracy compared to the common lightweight solutions, and significant footprint reduction of backbone inference time (up to 100x) and GPU consumption (up to 10x) compared to the heavy transformer-based solutions.

Edward Fish, Umberto Michieli, Mete Ozay

Recent advancement in Automatic Speech Recognition (ASR) has produced large AI models, which become impractical for deployment in mobile devices. Model quantization is effective to produce compressed general-purpose models, however such models may only be deployed to a restricted sub-domain of interest. We show that ASR models can be personalized during quantization while relying on just a small set of unlabelled samples from the target domain. To this end, we propose myQASR, a mixed-precision quantization method that generates tailored quantization schemes for diverse users under any memory requirement with no fine-tuning. myQASR automatically evaluates the quantization sensitivity of network layers by analysing the full-precision activation values. We are then able to generate a personalised mixed-precision quantization scheme for any pre-determined memory budget. Results for large-scale ASR models show how myQASR improves performance for specific genders, languages, and speakers.

Francesco Barbato, Umberto Michieli, Jijoong Moon et al.

Recent years have seen object detection robotic systems deployed in several personal devices (e.g., home robots and appliances). This has highlighted a challenge in their design, i.e., they cannot efficiently update their knowledge to distinguish between general classes and user-specific instances (e.g., a dog vs. user's dog). We refer to this challenging task as Instance-level Personalized Object Detection (IPOD). The personalization task requires many samples for model tuning and optimization in a centralized server, raising privacy concerns. An alternative is provided by approaches based on recent large-scale Foundation Models, but their compute costs preclude on-device applications. In our work we tackle both problems at the same time, designing a Few-Shot IPOD strategy called AuXFT. We introduce a conditional coarse-to-fine few-shot learner to refine the coarse predictions made by an efficient object detector, showing that using an off-the-shelf model leads to poor personalization due to neural collapse. Therefore, we introduce a Translator block that generates an auxiliary feature space where features generated by a self-supervised model (e.g., DINOv2) are distilled without impacting the performance of the detector. We validate AuXFT on three publicly available datasets and one in-house benchmark designed for the IPOD task, achieving remarkable gains in all considered scenarios with excellent time-complexity trade-off: AuXFT reaches a performance of 80% its upper bound at just 32% of the inference time, 13% of VRAM and 19% of the model size.

Massimo Bini, Ondrej Bohdal, Umberto Michieli et al.

Memory-augmented Large Language Models (LLMs) have demonstrated remarkable consistency during prolonged dialogues by storing relevant memories and incorporating them as context. Such memory-based personalization is also key in on-device settings that allow users to keep their conversations and data private. However, memory-augmented systems typically rely on LLMs that are too costly for local on-device deployment. Even though Small Language Models (SLMs) are more suitable for on-device inference than LLMs, they cannot achieve sufficient performance. Additionally, these LLM-based systems lack native visual capabilities, limiting their applicability in multimodal contexts. In this paper, we introduce (i) MemLoRA, a novel memory system that enables local deployment by equipping SLMs with specialized memory adapters, and (ii) its vision extension MemLoRA-V, which integrates small Vision-Language Models (SVLMs) to memory systems, enabling native visual understanding. Following knowledge distillation principles, each adapter is trained separately for specific memory operations$\unicode{x2013}$knowledge extraction, memory update, and memory-augmented generation. Equipped with memory adapters, small models enable accurate on-device memory operations without cloud dependency. On text-only operations, MemLoRA outperforms 10$\times$ larger baseline models (e.g., Gemma2-27B) and achieves performance comparable to 60$\times$ larger models (e.g., GPT-OSS-120B) on the LoCoMo benchmark. To evaluate visual understanding operations instead, we extend LoCoMo with challenging Visual Question Answering tasks that require direct visual reasoning. On this, our VLM-integrated MemLoRA-V shows massive improvements over caption-based approaches (81.3 vs. 23.7 accuracy) while keeping strong performance in text-based tasks, demonstrating the efficacy of our method in multimodal contexts.

Kirill Paramonov, Mete Ozay, Aristeidis Mystakidis et al.

The proliferation of AI models in everyday devices has highlighted a critical challenge: prediction errors that degrade user experience. While existing solutions focus on error detection, they rarely provide efficient correction mechanisms, especially for resource-constrained devices. We present a novel system enabling users to correct AI misclassifications through few-shot learning, requiring minimal computational resources and storage. Our approach combines server-side foundation model training with on-device prototype-based classification, enabling efficient error correction through prototype updates rather than model retraining. The system consists of two key components: (1) a server-side pipeline that leverages knowledge distillation to transfer robust feature representations from foundation models to device-compatible architectures, and (2) a device-side mechanism that enables ultra-efficient error correction through prototype adaptation. We demonstrate our system's effectiveness on both image classification and object detection tasks, achieving over 50% error correction in one-shot scenarios on Food-101 and Flowers-102 datasets while maintaining minimal forgetting (less than 0.02%) and negligible computational overhead. Our implementation, validated through an Android demonstration app, proves the system's practicality in real-world scenarios.

Haaris Mehmood, Agnieszka Dobrowolska, Karthikeyan Saravanan et al.

Federated Learning (FL) enables training state-of-the-art Automatic Speech Recognition (ASR) models on user devices (clients) in distributed systems, hence preventing transmission of raw user data to a central server. A key challenge facing practical adoption of FL for ASR is obtaining ground-truth labels on the clients. Existing approaches rely on clients to manually transcribe their speech, which is impractical for obtaining large training corpora. A promising alternative is using semi-/self-supervised learning approaches to leverage unlabelled user data. To this end, we propose FedNST, a novel method for training distributed ASR models using private and unlabelled user data. We explore various facets of FedNST, such as training models with different proportions of labelled and unlabelled data, and evaluate the proposed approach on 1173 simulated clients. Evaluating FedNST on LibriSpeech, where 960 hours of speech data is split equally into server (labelled) and client (unlabelled) data, showed a 22.5% relative word error rate reduction} (WERR) over a supervised baseline trained only on server data.

Elena Camuffo, Umberto Michieli, Simone Milani et al.

Developing a reliable vision system is a fundamental challenge for robotic technologies (e.g., indoor service robots and outdoor autonomous robots) which can ensure reliable navigation even in challenging environments such as adverse weather conditions (e.g., fog, rain), poor lighting conditions (e.g., over/under exposure), or sensor degradation (e.g., blurring, noise), and can guarantee high performance in safety-critical functions. Current solutions proposed to improve model robustness usually rely on generic data augmentation techniques or employ costly test-time adaptation methods. In addition, most approaches focus on addressing a single vision task (typically, image recognition) utilising synthetic data. In this paper, we introduce Per-corruption Adaptation of Normalization statistics (PAN) to enhance the model robustness of vision systems. Our approach entails three key components: (i) a corruption type identification module, (ii) dynamic adjustment of normalization layer statistics based on identified corruption type, and (iii) real-time update of these statistics according to input data. PAN can integrate seamlessly with any convolutional model for enhanced accuracy in several robot vision tasks. In our experiments, PAN obtains robust performance improvement on challenging real-world corrupted image datasets (e.g., OpenLoris, ExDark, ACDC), where most of the current solutions tend to fail. Moreover, PAN outperforms the baseline models by 20-30% on synthetic benchmarks in object recognition tasks.

Yan Zhang, Mete Ozay, Shuohao Li et al.

Recent study shows that a wide deep network can obtain accuracy comparable to a deeper but narrower network. Compared to narrower and deeper networks, wide networks employ relatively less number of layers and have various important benefits, such that they have less running time on parallel computing devices, and they are less affected by gradient vanishing problems. However, the parameter size of a wide network can be very large due to use of large width of each layer in the network. In order to keep the benefits of wide networks meanwhile improve the parameter size and accuracy trade-off of wide networks, we propose a binary tree architecture to truncate architecture of wide networks by reducing the width of the networks. More precisely, in the proposed architecture, the width is continuously reduced from lower layers to higher layers in order to increase the expressive capacity of network with a less increase on parameter size. Also, to ease the gradient vanishing problem, features obtained at different layers are concatenated to form the output of our architecture. By employing the proposed architecture on a baseline wide network, we can construct and train a new network with same depth but considerably less number of parameters. In our experimental analyses, we observe that the proposed architecture enables us to obtain better parameter size and accuracy trade-off compared to baseline networks using various benchmark image classification datasets. The results show that our model can decrease the classification error of baseline from 20.43% to 19.22% on Cifar-100 using only 28% of parameters that baseline has. Code is available at https://github.com/ZhangVision/bitnet.

Zhun Sun, Mete Ozay, Takayuki Okatani

Despite the effectiveness of Convolutional Neural Networks (CNNs) for image classification, our understanding of the relationship between shape of convolution kernels and learned representations is limited. In this work, we explore and employ the relationship between shape of kernels which define Receptive Fields (RFs) in CNNs for learning of feature representations and image classification. For this purpose, we first propose a feature visualization method for visualization of pixel-wise classification score maps of learned features. Motivated by our experimental results, and observations reported in the literature for modeling of visual systems, we propose a novel design of shape of kernels for learning of representations in CNNs. In the experimental results, we achieved a state-of-the-art classification performance compared to a base CNN model [28] by reducing the number of parameters and computational time of the model using the ILSVRC-2012 dataset [24]. The proposed models also outperform the state-of-the-art models employed on the CIFAR-10/100 datasets [12] for image classification. Additionally, we analyzed the robustness of the proposed method to occlusion for classification of partially occluded images compared with the state-of-the-art methods. Our results indicate the effectiveness of the proposed approach. The code is available in github.com/minogame/caffe-qhconv.

Juan Borrego-Carazo, Mete Ozay, Frederik Laboyrie et al.

Recovering a high dynamic range (HDR) image from a single low dynamic range (LDR) image, namely inverse tone mapping (ITM), is challenging due to the lack of information in over- and under-exposed regions. Current methods focus exclusively on training high-performing but computationally inefficient ITM models, which in turn hinder deployment of the ITM models in resource-constrained environments with limited computing power such as edge and mobile device applications. To this end, we propose combining efficient operations of deep neural networks with a novel mixed quantization scheme to construct a well-performing but computationally efficient mixed quantization network (MQN) which can perform single image ITM on mobile platforms. In the ablation studies, we explore the effect of using different attention mechanisms, quantization schemes, and loss functions on the performance of MQN in ITM tasks. In the comparative analyses, ITM models trained using MQN perform on par with the state-of-the-art methods on benchmark datasets. MQN models provide up to 10 times improvement on latency and 25 times improvement on memory consumption.

Jie Xu, Karthikeyan Saravanan, Rogier van Dalen et al. · cambridge

Federated learning (FL) allows clients to collaboratively train a global model without sharing their local data with a server. However, clients' contributions to the server can still leak sensitive information. Differential privacy (DP) addresses such leakage by providing formal privacy guarantees, with mechanisms that add randomness to the clients' contributions. The randomness makes it infeasible to train large transformer-based models, common in modern federated learning systems. In this work, we empirically evaluate the practicality of fine-tuning large scale on-device transformer-based models with differential privacy in a federated learning system. We conduct comprehensive experiments on various system properties for tasks spanning a multitude of domains: speech recognition, computer vision (CV) and natural language understanding (NLU). Our results show that full fine-tuning under differentially private federated learning (DP-FL) generally leads to huge performance degradation which can be alleviated by reducing the dimensionality of contributions through parameter-efficient fine-tuning (PEFT). Our benchmarks of existing DP-PEFT methods show that DP-Low-Rank Adaptation (DP-LoRA) consistently outperforms other methods. An even more promising approach, DyLoRA, which makes the low rank variable, when naively combined with FL would straightforwardly break differential privacy. We therefore propose an adaptation method that can be combined with differential privacy and call it DP-DyLoRA. Finally, we are able to reduce the accuracy degradation and word error rate (WER) increase due to DP to less than 2% and 7% respectively with 1 million clients and a stringent privacy budget of $ε=2$.

Donald Shenaj, Ondrej Bohdal, Mete Ozay et al.

Recent advancements in image generation models have enabled personalized image creation with both user-defined subjects (content) and styles. Prior works achieved personalization by merging corresponding low-rank adapters (LoRAs) through optimization-based methods, which are computationally demanding and unsuitable for real-time use on resource-constrained devices like smartphones. To address this, we introduce LoRA$.$rar, a method that not only improves image quality but also achieves a remarkable speedup of over $4000\times$ in the merging process. We collect a dataset of style and subject LoRAs and pre-train a hypernetwork on a diverse set of content-style LoRA pairs, learning an efficient merging strategy that generalizes to new, unseen content-style pairs, enabling fast, high-quality personalization. Moreover, we identify limitations in existing evaluation metrics for content-style quality and propose a new protocol using multimodal large language models (MLLMs) for more accurate assessment. Our method significantly outperforms the current state of the art in both content and style fidelity, as validated by MLLM assessments and human evaluations.

Emanuele Aiello, Umberto Michieli, Diego Valsesia et al.

Personalized image generation requires text-to-image generative models that capture the core features of a reference subject to allow for controlled generation across different contexts. Existing methods face challenges due to complex training requirements, high inference costs, limited flexibility, or a combination of these issues. In this paper, we introduce DreamCache, a scalable approach for efficient and high-quality personalized image generation. By caching a small number of reference image features from a subset of layers and a single timestep of the pretrained diffusion denoiser, DreamCache enables dynamic modulation of the generated image features through lightweight, trained conditioning adapters. DreamCache achieves state-of-the-art image and text alignment, utilizing an order of magnitude fewer extra parameters, and is both more computationally effective and versatile than existing models.

Umberto Michieli, Mete Ozay

Continual Learning (CL) aims to learn a sequence of problems (i.e., tasks and domains) by transferring knowledge acquired on previous problems, whilst avoiding forgetting of past ones. Different from previous approaches which focused on CL for one NLP task or domain in a specific use-case, in this paper, we address a more general CL setting to learn from a sequence of problems in a unique framework. Our method, HOP, permits to hop across tasks and domains by addressing the CL problem along three directions: (i) we employ a set of adapters to generalize a large pre-trained model to unseen problems, (ii) we compute high-order moments over the distribution of embedded representations to distinguish independent and correlated statistics across different tasks and domains, (iii) we process this enriched information with auxiliary heads specialized for each end problem. Extensive experimental campaign on 4 NLP applications, 5 benchmarks and 2 CL setups demonstrates the effectiveness of our HOP.

Taha Ceritli, Ondrej Bohdal, Mete Ozay et al.

Large language models (LLMs) often leverage adapters, such as low-rank-based adapters, to achieve strong performance on downstream tasks. However, storing a separate adapter for each task significantly increases memory requirements, posing a challenge for resource-constrained environments such as mobile devices. Although model merging techniques can reduce storage costs, they typically result in substantial performance degradation. In this work, we introduce HydraOpt, a new model merging technique that capitalizes on the inherent similarities between the matrices of low-rank adapters. Unlike existing methods that produce a fixed trade-off between storage size and performance, HydraOpt allows us to navigate this spectrum of efficiency and performance. Our experiments show that HydraOpt significantly reduces storage size (48% reduction) compared to storing all adapters, while achieving competitive performance (0.2-1.8% drop). Furthermore, it outperforms existing merging techniques in terms of performance at the same or slightly worse storage efficiency.

Umberto Michieli, Jijoong Moon, Daehyun Kim et al.

Nowadays, users demand for increased personalization of vision systems to localize and identify personal instances of objects (e.g., my dog rather than dog) from a few-shot dataset only. Despite outstanding results of deep networks on classical label-abundant benchmarks (e.g., those of the latest YOLOv8 model for standard object detection), they struggle to maintain within-class variability to represent different instances rather than object categories only. We construct an Object-conditioned Bag of Instances (OBoI) based on multi-order statistics of extracted features, where generic object detection models are extended to search and identify personal instances from the OBoI's metric space, without need for backpropagation. By relying on multi-order statistics, OBoI achieves consistent superior accuracy in distinguishing different instances. In the results, we achieve 77.1% personal object recognition accuracy in case of 18 personal instances, showing about 12% relative gain over the state of the art.

Elena Camuffo, Umberto Michieli, Jijoong Moon et al.

Improving model robustness in case of corrupted images is among the key challenges to enable robust vision systems on smart devices, such as robotic agents. Particularly, robust test-time performance is imperative for most of the applications. This paper presents a novel approach to improve robustness of any classification model, especially on severely corrupted images. Our method (FROST) employs high-frequency features to detect input image corruption type, and select layer-wise feature normalization statistics. FROST provides the state-of-the-art results for different models and datasets, outperforming competitors on ImageNet-C by up to 37.1% relative gain, improving baseline of 40.9% mCE on severe corruptions.

Hafiz Tiomoko Ali, Umberto Michieli, Ji Joong Moon et al.

The recently discovered Neural collapse (NC) phenomenon states that the last-layer weights of Deep Neural Networks (DNN), converge to the so-called Equiangular Tight Frame (ETF) simplex, at the terminal phase of their training. This ETF geometry is equivalent to vanishing within-class variability of the last layer activations. Inspired by NC properties, we explore in this paper the transferability of DNN models trained with their last layer weight fixed according to ETF. This enforces class separation by eliminating class covariance information, effectively providing implicit regularization. We show that DNN models trained with such a fixed classifier significantly improve transfer performance, particularly on out-of-domain datasets. On a broad range of fine-grained image classification datasets, our approach outperforms i) baseline methods that do not perform any covariance regularization (up to 22%), as well as ii) methods that explicitly whiten covariance of activations throughout training (up to 19%). Our findings suggest that DNNs trained with fixed ETF classifiers offer a powerful mechanism for improving transfer learning across domains.

Francesco Barbato, Umberto Michieli, Mehmet Kerim Yucel et al.

In multimedia understanding tasks, corrupted samples pose a critical challenge, because when fed to machine learning models they lead to performance degradation. In the past, three groups of approaches have been proposed to handle noisy data: i) enhancer and denoiser modules to improve the quality of the noisy data, ii) data augmentation approaches, and iii) domain adaptation strategies. All the aforementioned approaches come with drawbacks that limit their applicability; the first has high computational costs and requires pairs of clean-corrupted data for training, while the others only allow deployment of the same task/network they were trained on (\ie, when upstream and downstream task/network are the same). In this paper, we propose SyMPIE to solve these shortcomings. To this end, we design a small, modular, and efficient (just 2GFLOPs to process a Full HD image) system to enhance input data for robust downstream multimedia understanding with minimal computational cost. Our SyMPIE is pre-trained on an upstream task/network that should not match the downstream ones and does not need paired clean-corrupted samples. Our key insight is that most input corruptions found in real-world tasks can be modeled through global operations on color channels of images or spatial filters with small kernels. We validate our approach on multiple datasets and tasks, such as image classification (on ImageNetC, ImageNetC-Bar, VizWiz, and a newly proposed mixed corruption benchmark named ImageNetC-mixed) and semantic segmentation (on Cityscapes, ACDC, and DarkZurich) with consistent improvements of about 5\% relative accuracy gain across the board. The code of our approach and the new ImageNetC-mixed benchmark will be made available upon publication.

Kirill Paramonov, Mete Ozay, Eunju Yang et al.

Class-incremental learning in the context of limited personal labeled samples (few-shot) is critical for numerous real-world applications, such as smart home devices. A key challenge in these scenarios is balancing the trade-off between adapting to new, personalized classes and maintaining the performance of the model on the original, base classes. Fine-tuning the model on novel classes often leads to the phenomenon of catastrophic forgetting, where the accuracy of base classes declines unpredictably and significantly. In this paper, we propose a simple yet effective mechanism to address this challenge by controlling the trade-off between novel and base class accuracy. We specifically target the ultra-low-shot scenario, where only a single example is available per novel class. Our approach introduces a Novel Class Detection (NCD) rule, which adjusts the degree of forgetting a priori while simultaneously enhancing performance on novel classes. We demonstrate the versatility of our solution by applying it to state-of-the-art Few-Shot Class-Incremental Learning (FSCIL) methods, showing consistent improvements across different settings. To better quantify the trade-off between novel and base class performance, we introduce new metrics: NCR@2FOR and NCR@5FOR. Our approach achieves up to a 30% improvement in novel class accuracy on the CIFAR100 dataset (1-shot, 1 novel class) while maintaining a controlled base class forgetting rate of 2%.

Donald Shenaj, Ondrej Bohdal, Taha Ceritli et al.

On-device deployment of Large Language Models (LLMs) frequently leverages Low-Rank Adapters (LoRAs) to support diverse downstream tasks under tight resource constraints. To address the limited storage capacity of mobile devices, recent works have explored model merging techniques to fuse multiple LoRAs into a single one. In practice, however, LoRAs are often delivered incrementally, as users request support for new tasks (e.g., novel problem types or languages). This scenario introduces a new challenge: on-device online continual merging, where the objective is to incorporate new LoRAs while preserving the performance on previously supported tasks. In this paper, we propose a data-free and computationally efficient strategy for selecting and merging LoRAs when a new one becomes available, assuming the device can store only a limited number of adapters. Extensive experiments across real-world tasks demonstrate the superiority of our approach compared to alternative strategies while adhering to the storage budget and compute limitations of on-device settings.

Junyi Zhu, Savas Ozkan, Andrea Maracani et al.

Deploying natural language processing (NLP) models on mobile platforms requires models that can adapt across diverse applications while remaining efficient in memory and computation. We investigate pre-finetuning strategies to enhance the adaptability of lightweight BERT-like encoders for two fundamental NLP task families: named entity recognition (NER) and text classification. While pre-finetuning improves downstream performance for each task family individually, we find that naïve multi-task pre-finetuning introduces conflicting optimization signals that degrade overall performance. To address this, we propose a simple yet effective multi-task pre-finetuning framework based on task-primary LoRA modules, which enables a single shared encoder backbone with modular adapters. Our approach achieves performance comparable to individual pre-finetuning while meeting practical deployment constraint. Experiments on 21 downstream tasks show average improvements of +0.8% for NER and +8.8% for text classification, demonstrating the effectiveness of our method for versatile mobile NLP applications.

Elena Camuffo, Francesco Barbato, Mete Ozay et al.

We introduce MOCHA (Multi-modal Objects-aware Cross-arcHitecture Alignment), a knowledge distillation approach that transfers region-level multimodal semantics from a large vision-language teacher (e.g., LLaVa) into a lightweight vision-only object detector student (e.g., YOLO). A translation module maps student features into a joint space, where the training of the student and translator is guided by a dual-objective loss that enforces both local alignment and global relational consistency. Unlike prior approaches focused on dense or global alignment, MOCHA operates at the object level, enabling efficient transfer of semantics without modifying the teacher or requiring textual input at inference. We validate our method across four personalized detection benchmarks under few-shot regimes. Results show consistent gains over baselines, with a +10.1 average score improvement. Despite its compact architecture, MOCHA reaches performance on par with larger multimodal models, proving its suitability for real-world deployment.

Ondrej Bohdal, Mete Ozay, Jijoong Moon et al.

Adapter parameters provide a mechanism to modify the behavior of machine learning models and have gained significant popularity in the context of large language models (LLMs) and generative AI. These parameters can be merged to support multiple tasks via a process known as task merging. However, prior work on merging in LLMs, particularly in natural language processing, has been limited to scenarios where each test example addresses only a single task. In this paper, we focus on on-device settings and study the problem of text-based compositional multi-tasking, where each test example involves the simultaneous execution of multiple tasks. For instance, generating a translated summary of a long text requires solving both translation and summarization tasks concurrently. To facilitate research in this setting, we propose a benchmark comprising four practically relevant compositional tasks. We also present an efficient method (Learnable Calibration) tailored for on-device applications, where computational resources are limited, emphasizing the need for solutions that are both resource-efficient and high-performing. Our contributions lay the groundwork for advancing the capabilities of LLMs in real-world multi-tasking scenarios, expanding their applicability to complex, resource-constrained use cases.

Junyi Zhu, Ruicong Yao, Taha Ceritli et al.

Current network training paradigms primarily focus on either centralized or decentralized data regimes. However, in practice, data availability often exhibits a hybrid nature, where both regimes coexist. This hybrid setting presents new opportunities for model training, as the two regimes offer complementary trade-offs: decentralized data is abundant but subject to heterogeneity and communication constraints, while centralized data, though limited in volume and potentially unrepresentative, enables better curation and high-throughput access. Despite its potential, effectively combining these paradigms remains challenging, and few frameworks are tailored to hybrid data regimes. To address this, we propose a novel framework that constructs a model atlas from decentralized models and leverages centralized data to refine a global model within this structured space. The refined model is then used to reinitialize the decentralized models. Our method synergizes federated learning (to exploit decentralized data) and model merging (to utilize centralized data), enabling effective training under hybrid data availability. Theoretically, we show that our approach achieves faster convergence than methods relying solely on decentralized data, due to variance reduction in the merging process. Extensive experiments demonstrate that our framework consistently outperforms purely centralized, purely decentralized, and existing hybrid-adaptable methods. Notably, our method remains robust even when the centralized and decentralized data domains differ or when decentralized data contains noise, significantly broadening its applicability.

Hasan Abed Al Kader Hammoud, Umberto Michieli, Fabio Pizzati et al.

Merging Large Language Models (LLMs) is a cost-effective technique for combining multiple expert LLMs into a single versatile model, retaining the expertise of the original ones. However, current approaches often overlook the importance of safety alignment during merging, leading to highly misaligned models. This work investigates the effects of model merging on alignment. We evaluate several popular model merging techniques, demonstrating that existing methods do not only transfer domain expertise but also propagate misalignment. We propose a simple two-step approach to address this problem: (i) generating synthetic safety and domain-specific data, and (ii) incorporating these generated data into the optimization process of existing data-aware model merging techniques. This allows us to treat alignment as a skill that can be maximized in the resulting merged LLM. Our experiments illustrate the effectiveness of integrating alignment-related data during merging, resulting in models that excel in both domain expertise and alignment.

Jiwei Tian, Buhong Wang, Jing Li et al.

Vulnerability of various machine learning methods to adversarial examples has been recently explored in the literature. Power systems which use these vulnerable methods face a huge threat against adversarial examples. To this end, we first propose a signal-specific method and a universal signal-agnostic method to attack power systems using generated adversarial examples. Black-box attacks based on transferable characteristics and the above two methods are also proposed and evaluated. We then adopt adversarial training to defend systems against adversarial attacks. Experimental analyses demonstrate that our signal-specific attack method provides less perturbation compared to the FGSM (Fast Gradient Sign Method), and our signal-agnostic attack method can generate perturbations fooling most natural signals with high probability. What's more, the attack method based on the universal signal-agnostic algorithm has a higher transfer rate of black-box attacks than the attack method based on the signal-specific algorithm. In addition, the results show that the proposed adversarial training improves robustness of power systems to adversarial examples.

Shuang Liu, Mete Ozay

Zero-shot domain adaptation (ZDA) methods aim to transfer knowledge about a task learned in a source domain to a target domain, while data from target domain are not available. In this work, we address learning feature representations which are invariant to and shared among different domains considering task characteristics for ZDA. To this end, we propose a method for task-guided ZDA (TG-ZDA) which employs multi-branch deep neural networks to learn feature representations exploiting their domain invariance and shareability properties. The proposed TG-ZDA models can be trained end-to-end without requiring synthetic tasks and data generated from estimated representations of target domains. The proposed TG-ZDA has been examined using benchmark ZDA tasks on image classification datasets. Experimental results show that our proposed TG-ZDA outperforms state-of-the-art ZDA methods for different domains and tasks.

Umberto Michieli, Mete Ozay

Federated Learning (FL) is a framework which enables distributed model training using a large corpus of decentralized training data. Existing methods aggregate models disregarding their internal representations, which are crucial for training models in vision tasks. System and statistical heterogeneity (e.g., highly imbalanced and non-i.i.d. data) further harm model training. To this end, we introduce a method, called FedProto, which computes client deviations using margins of prototypical representations learned on distributed data, and applies them to drive federated optimization via an attention mechanism. In addition, we propose three methods to analyse statistical properties of feature representations learned in FL, in order to elucidate the relationship between accuracy, margins and feature discrepancy of FL models. In experimental analyses, FedProto demonstrates state-of-the-art accuracy and convergence rate across image classification and semantic segmentation benchmarks by enabling maximum margin training of FL models. Moreover, FedProto reduces uncertainty of predictions of FL models compared to the baseline. To our knowledge, this is the first work evaluating FL models in dense prediction tasks, such as semantic segmentation.

Piotr Kicki, Mete Ozay, Piotr Skrzypczyński

We propose a computationally efficient $G$-invariant neural network that approximates functions invariant to the action of a given permutation subgroup $G \leq S_n$ of the symmetric group on input data. The key element of the proposed network architecture is a new $G$-invariant transformation module, which produces a $G$-invariant latent representation of the input data. Theoretical considerations are supported by numerical experiments, which demonstrate the effectiveness and strong generalization properties of the proposed method in comparison to other $G$-invariant neural networks.

Piotr Kicki, Tomasz Gawron, Krzysztof Ćwian et al.

Being able to rapidly respond to the changing scenes and traffic situations by generating feasible local paths is of pivotal importance for car autonomy. We propose to train a deep neural network (DNN) to plan feasible and nearly-optimal paths for kinematically constrained vehicles in small constant time. Our DNN model is trained using a novel weakly supervised approach and a gradient-based policy search. On real and simulated scenes and a large set of local planning problems, we demonstrate that our approach outperforms the existing planners with respect to the number of successfully completed tasks. While the path generation time is about 40 ms, the generated paths are smooth and comparable to those obtained from conventional path planners.

Piotr Kicki, Mete Ozay, Piotr Skrzypczyński

We introduce a method to design a computationally efficient $G$-invariant neural network that approximates functions invariant to the action of a given permutation subgroup $G \leq S_n$ of the symmetric group on input data. The key element of the proposed network architecture is a new $G$-invariant transformation module, which produces a $G$-invariant latent representation of the input data. This latent representation is then processed with a multi-layer perceptron in the network. We prove the universality of the proposed architecture, discuss its properties and highlight its computational and memory efficiency. Theoretical considerations are supported by numerical experiments involving different network configurations, which demonstrate the effectiveness and strong generalization properties of the proposed method in comparison to other $G$-invariant neural networks.

Mete Ozay

In recent studies, several asymptotic upper bounds on generalization errors on deep neural networks (DNNs) are theoretically derived. These bounds are functions of several norms of weights of the DNNs, such as the Frobenius and spectral norms, and they are computed for weights grouped according to either input and output channels of the DNNs. In this work, we conjecture that if we can impose multiple constraints on weights of DNNs to upper bound the norms of the weights, and train the DNNs with these weights, then we can attain empirical generalization errors closer to the derived theoretical bounds, and improve accuracy of the DNNs. To this end, we pose two problems. First, we aim to obtain weights whose different norms are all upper bounded by a constant number, e.g. 1.0. To achieve these bounds, we propose a two-stage renormalization procedure; (i) normalization of weights according to different norms used in the bounds, and (ii) reparameterization of the normalized weights to set a constant and finite upper bound of their norms. In the second problem, we consider training DNNs with these renormalized weights. To this end, we first propose a strategy to construct joint spaces (manifolds) of weights according to different constraints in DNNs. Next, we propose a fine-grained SGD algorithm (FG-SGD) for optimization on the weight manifolds to train DNNs with assurance of convergence to minima. Experimental results show that image classification accuracy of baseline DNNs can be boosted using FG-SGD on collections of manifolds identified by multiple constraints.

Mingzhen Shao, Zhun Sun, Mete Ozay et al.

We address a problem of estimating pose of a person's head from its RGB image. The employment of CNNs for the problem has contributed to significant improvement in accuracy in recent works. However, we show that the following two methods, despite their simplicity, can attain further improvement: (i) proper adjustment of the margin of bounding box of a detected face, and (ii) choice of loss functions. We show that the integration of these two methods achieve the new state-of-the-art on standard benchmark datasets for in-the-wild head pose estimation.

Junjie Hu, Mete Ozay, Yan Zhang et al.

This paper considers the problem of single image depth estimation. The employment of convolutional neural networks (CNNs) has recently brought about significant advancements in the research of this problem. However, most existing methods suffer from loss of spatial resolution in the estimated depth maps; a typical symptom is distorted and blurry reconstruction of object boundaries. In this paper, toward more accurate estimation with a focus on depth maps with higher spatial resolution, we propose two improvements to existing approaches. One is about the strategy of fusing features extracted at different scales, for which we propose an improved network architecture consisting of four modules: an encoder, decoder, multi-scale feature fusion module, and refinement module. The other is about loss functions for measuring inference errors used in training. We show that three loss terms, which measure errors in depth, gradients and surface normals, respectively, contribute to improvement of accuracy in an complementary fashion. Experimental results show that these two improvements enable to attain higher accuracy than the current state-of-the-arts, which is given by finer resolution reconstruction, for example, with small objects and object boundaries.

Masanori Suganuma, Mete Ozay, Takayuki Okatani

Researchers have applied deep neural networks to image restoration tasks, in which they proposed various network architectures, loss functions, and training methods. In particular, adversarial training, which is employed in recent studies, seems to be a key ingredient to success. In this paper, we show that simple convolutional autoencoders (CAEs) built upon only standard network components, i.e., convolutional layers and skip connections, can outperform the state-of-the-art methods which employ adversarial training and sophisticated loss functions. The secret is to employ an evolutionary algorithm to automatically search for good architectures. Training optimized CAEs by minimizing the $\ell_2$ loss between reconstructed images and their ground truths using the ADAM optimizer is all we need. Our experimental results show that this approach achieves 27.8 dB peak signal to noise ratio (PSNR) on the CelebA dataset and 40.4 dB on the SVHN dataset, compared to 22.8 dB and 33.0 dB provided by the former state-of-the-art methods, respectively.

Fazil Altinel, Mete Ozay, Takayuki Okatani

In this paper, we propose a structured image inpainting method employing an energy based model. In order to learn structural relationship between patterns observed in images and missing regions of the images, we employ an energy-based structured prediction method. The structural relationship is learned by minimizing an energy function which is defined by a simple convolutional neural network. The experimental results on various benchmark datasets show that our proposed method significantly outperforms the state-of-the-art methods which use Generative Adversarial Networks (GANs). We obtained 497.35 mean squared error (MSE) on the Olivetti face dataset compared to 833.0 MSE provided by the state-of-the-art method. Moreover, we obtained 28.4 dB peak signal to noise ratio (PSNR) on the SVHN dataset and 23.53 dB on the CelebA dataset, compared to 22.3 dB and 21.3 dB, provided by the state-of-the-art methods, respectively. The code is publicly available.

Shuang Liu, Mete Ozay, Takayuki Okatani et al.

Autonomous underwater vehicles (AUVs) have been deployed for underwater exploration. However, its potential is confined by its limited on-board battery energy and data storage capacity. This problem has been addressed using docking systems by underwater recharging and data transfer for AUVs. In this work, we propose a vision based framework for underwater docking following these systems. The proposed framework comprises two modules; (i) a detection module which provides location information on underwater docking stations in 2D images captured by an on-board camera, and (ii) a pose estimation module which recovers the relative 3D position and orientation between docking stations and AUVs from the 2D images. For robust and credible detection of docking stations, we propose a convolutional neural network called Docking Neural Network (DoNN). For accurate pose estimation, a perspective-n-point algorithm is integrated into our framework. In order to examine our framework in underwater docking tasks, we collected a dataset of 2D images, named Underwater Docking Images Dataset (UDID), in an experimental water pool. To the best of our knowledge, UDID is the first publicly available underwater docking dataset. In the experiments, we first evaluate performance of the proposed detection module on UDID and its deformed variations. Next, we assess the accuracy of the pose estimation module by ground experiments, since it is not feasible to obtain true relative position and orientation between docking stations and AUVs under water. Then, we examine the pose estimation module by underwater experiments in our experimental water pool. Experimental results show that the proposed framework can be used to detect docking stations and estimate their relative pose efficiently and successfully, compared to the state-of-the-art baseline systems.

Zhun Sun, Mete Ozay, Takayuki Okatani

Deep learning algorithms have been known to be vulnerable to adversarial perturbations in various tasks such as image classification. This problem was addressed by employing several defense methods for detection and rejection of particular types of attacks. However, training and manipulating networks according to particular defense schemes increases computational complexity of the learning algorithms. In this work, we propose a simple yet effective method to improve robustness of convolutional neural networks (CNNs) to adversarial attacks by using data dependent adaptive convolution kernels. To this end, we propose a new type of HyperNetwork in order to employ statistical properties of input data and features for computation of statistical adaptive maps. Then, we filter convolution weights of CNNs with the learned statistical maps to compute dynamic kernels. Thereby, weights and kernels are collectively optimized for learning of image classification models robust to adversarial attacks without employment of additional target detection and rejection algorithms. We empirically demonstrate that the proposed method enables CNNs to spontaneously defend against different types of attacks, e.g. attacks generated by Gaussian noise, fast gradient sign methods (Goodfellow et al., 2014) and a black-box attack(Narodytska & Kasiviswanathan, 2016).

Zhun Sun, Mete Ozay, Takayuki Okatani

We develop a novel method for training of GANs for unsupervised and class conditional generation of images, called Linear Discriminant GAN (LD-GAN). The discriminator of an LD-GAN is trained to maximize the linear separability between distributions of hidden representations of generated and targeted samples, while the generator is updated based on the decision hyper-planes computed by performing LDA over the hidden representations. LD-GAN provides a concrete metric of separation capacity for the discriminator, and we experimentally show that it is possible to stabilize the training of LD-GAN simply by calibrating the update frequencies between generators and discriminators in the unsupervised case, without employment of normalization methods and constraints on weights. In the class conditional generation tasks, the proposed method shows improved training stability together with better generalization performance compared to WGAN that employs an auxiliary classifier.

Zhun Sun, Mete Ozay, Takayuki Okatani

In this work, we address the problem of improvement of robustness of feature representations learned using convolutional neural networks (CNNs) to image deformation. We argue that higher moment statistics of feature distributions could be shifted due to image deformations, and the shift leads to degrade of performance and cannot be reduced by ordinary normalization methods as observed in experimental analyses. In order to attenuate this effect, we apply additional non-linearity in CNNs by combining power functions with learnable parameters into convolution operation. In the experiments, we observe that CNNs which employ the proposed method obtain remarkable boost in both the generalization performance and the robustness under various types of deformations using large scale benchmark datasets. For instance, a model equipped with the proposed method obtains 3.3\% performance boost in mAP on Pascal Voc object detection task using deformed images, compared to the reference model, while both models provide the same performance using original images. To the best of our knowledge, this is the first work that studies robustness of deep features learned using CNNs to a wide range of deformations for object recognition and detection.

Yan Zhang, Mete Ozay, Zhun Sun et al.

In this paper, we suggest a framework to make use of mutual information as a regularization criterion to train Auto-Encoders (AEs). In the proposed framework, AEs are regularized by minimization of the mutual information between input and encoding variables of AEs during the training phase. In order to estimate the entropy of the encoding variables and the mutual information, we propose a non-parametric method. We also give an information theoretic view of Variational AEs (VAEs), which suggests that VAEs can be considered as parametric methods that estimate entropy. Experimental results show that the proposed non-parametric models have more degree of freedom in terms of representation learning of features drawn from complex distributions such as Mixture of Gaussians, compared to methods which estimate entropy using parametric approaches, such as Variational AEs.

Mete Ozay, Takayuki Okatani

Recent advances in optimization methods used for training convolutional neural networks (CNNs) with kernels, which are normalized according to particular constraints, have shown remarkable success. This work introduces an approach for training CNNs using ensembles of joint spaces of kernels constructed using different constraints. For this purpose, we address a problem of optimization on ensembles of products of submanifolds (PEMs) of convolution kernels. To this end, we first propose three strategies to construct ensembles of PEMs in CNNs. Next, we expound their geometric properties (metric and curvature properties) in CNNs. We make use of our theoretical results by developing a geometry-aware SGD algorithm (G-SGD) for optimization on ensembles of PEMs to train CNNs. Moreover, we analyze convergence properties of G-SGD considering geometric properties of PEMs. In the experimental analyses, we employ G-SGD to train CNNs on Cifar-10, Cifar-100 and Imagenet datasets. The results show that geometric adaptive step size computation methods of G-SGD can improve training loss and convergence properties of CNNs. Moreover, we observe that classification performance of baseline CNNs can be boosted using G-SGD on ensembles of PEMs identified by multiple constraints.

Mete Ozay, Takayuki Okatani

Kernel normalization methods have been employed to improve robustness of optimization methods to reparametrization of convolution kernels, covariate shift, and to accelerate training of Convolutional Neural Networks (CNNs). However, our understanding of theoretical properties of these methods has lagged behind their success in applications. We develop a geometric framework to elucidate underlying mechanisms of a diverse range of kernel normalization methods. Our framework enables us to expound and identify geometry of space of normalized kernels. We analyze and delineate how state-of-the-art kernel normalization methods affect the geometry of search spaces of the stochastic gradient descent (SGD) algorithms in CNNs. Following our theoretical results, we propose a SGD algorithm with assurance of almost sure convergence of the methods to a solution at single minimum of classification loss of CNNs. Experimental results show that the proposed method achieves state-of-the-art performance for major image classification benchmarks with CNNs.