Alberto Presta

Alberto Presta, Grzegorz Stefanski, Michal Byra et al.

Multimodal deep neural networks enhance deep comprehension by integrating diverse data modalities. Data from different modalities are typically projected into a shared latent space for similarity computation, but this process is resource intensive due to large image encoders and equal processing of test data during prediction. Early exit methods reduce computational load by utilizing intermediate layers, saving time and memory. However, developing such methods is challenging for multimodal data like image-text pairs. This study investigates the semantic content distributions present in intermediate layers of encoders such as CLIP, which can be derived from textual descriptions. We introduce Text-Guided Exit Modules (T-GEMs) and a rate-based regularizer to control encoder usage costs while maintaining cross-modal understanding performance.

Michal Byra, Pawel Olszowiec, Grzegorz Stefanski et al.

Vision Transformers (ViTs) are built by stacking independently parameterized blocks, but it remains unclear how much of this depth requires layer specific transformations and how much can be realized through recurrent computation. We study this question with bViT, a single-block recurrent ViT in which one transformer block is applied repeatedly to process an image. This architecture preserves the iterative structure of a deep ViT while removing layer specific block parameterization, providing a controlled setting for studying recurrence in vision. On ImageNet-1K, a 12-step bViT-B achieves accuracy comparable to standard ViT-B under the same training recipe and computational budget, while using an order of magnitude fewer parameters. We observe that recurrent performance improves with representation width, with wider bViTs recovering much more of the performance of standard ViTs than narrow variants. We interpret this behavior as implicit depth multiplexing, where a shared block expresses multiple step-dependent computations through the evolving hidden state. Beyond ImageNet classification, bViT transfers competitively to downstream tasks and enables parameter-efficient fine-tuning. Mechanistic analyses of activations, attention and step-specific pruning show that the shared block changes its effective behavior across recurrent steps rather than simply repeating the same computation. Our results suggest that a large fraction of ViT depth can be implemented through recurrent reuse, provided that the representation space is sufficiently wide.

Gabriele Spadaro, Alberto Presta, Jhony H. Giraldo et al.

Efficient compression of low-bit-rate point clouds is critical for bandwidth-constrained applications. However, existing techniques mainly focus on high-fidelity reconstruction, requiring many bits for compression. This paper proposes a "Denoising Diffusion Probabilistic Model" (DDPM) architecture for point cloud compression (DDPM-PCC) at low bit-rates. A PointNet encoder produces the condition vector for the generation, which is then quantized via a learnable vector quantizer. This configuration allows to achieve a low bitrates while preserving quality. Experiments on ShapeNet and ModelNet40 show improved rate-distortion at low rates compared to standardized and state-of-the-art approaches. We publicly released the code at https://github.com/EIDOSLAB/DDPM-PCC.

Grzegorz Stefanski, Alberto Presta, Michal Byra

In pruning, the Lottery Ticket Hypothesis posits that large networks contain sparse subnetworks, or winning tickets, that can be trained in isolation to match the performance of their dense counterparts. However, most existing approaches assume a single universal winning ticket shared across all inputs, ignoring the inherent heterogeneity of real-world data. In this work, we propose Routing the Lottery (RTL), an adaptive pruning framework that discovers multiple specialized subnetworks, called adaptive tickets, each tailored to a class, semantic cluster, or environmental condition. Across diverse datasets and tasks, RTL consistently outperforms single- and multi-model baselines in balanced accuracy and recall, while using up to 10 times fewer parameters than independent models and exhibiting semantically aligned. Furthermore, we identify subnetwork collapse, a performance drop under aggressive pruning, and introduce a subnetwork similarity score that enables label-free diagnosis of oversparsification. Overall, our results recast pruning as a mechanism for aligning model structure with data heterogeneity, paving the way toward more modular and context-aware deep learning.

Alberto Presta, Gabriele Spadaro, Enzo Tartaglione et al.

In Learned Image Compression (LIC), a model is trained at encoding and decoding images sampled from a source domain, often outperforming traditional codecs on natural images; yet its performance may be far from optimal on images sampled from different domains. In this work, we tackle the problem of adapting a pre-trained model to multiple target domains by plugging into the decoder an adapter module for each of them, including the source one. Each adapter improves the decoder performance on a specific domain, without the model forgetting about the images seen at training time. A gate network computes the weights to optimally blend the contributions from the adapters when the bitstream is decoded. We experimentally validate our method over two state-of-the-art pre-trained models, observing improved rate-distortion efficiency on the target domains without penalties on the source domain. Furthermore, the gate's ability to find similarities with the learned target domains enables better encoding efficiency also for images outside them.

Guglielmo Gallone, Francesco Iodice, Alberto Presta et al.

Aims. To develop a deep-learning based system for recognition of subclinical atherosclerosis on a plain frontal chest x-ray. Methods and Results. A deep-learning algorithm to predict coronary artery calcium (CAC) score (the AI-CAC model) was developed on 460 chest x-ray (80% training cohort, 20% internal validation cohort) of primary prevention patients (58.4% male, median age 63 [51-74] years) with available paired chest x-ray and chest computed tomography (CT) indicated for any clinical reason and performed within 3 months. The CAC score calculated on chest CT was used as ground truth. The model was validated on an temporally-independent cohort of 90 patients from the same institution (external validation). The diagnostic accuracy of the AI-CAC model assessed by the area under the curve (AUC) was the primary outcome. Overall, median AI-CAC score was 35 (0-388) and 28.9% patients had no AI-CAC. AUC of the AI-CAC model to identify a CAC>0 was 0.90 in the internal validation cohort and 0.77 in the external validation cohort. Sensitivity was consistently above 92% in both cohorts. In the overall cohort (n=540), among patients with AI-CAC=0, a single ASCVD event occurred, after 4.3 years. Patients with AI-CAC>0 had significantly higher Kaplan Meier estimates for ASCVD events (13.5% vs. 3.4%, log-rank=0.013). Conclusion. The AI-CAC model seems to accurately detect subclinical atherosclerosis on chest x-ray with elevated sensitivity, and to predict ASCVD events with elevated negative predictive value. Adoption of the AI-CAC model to refine CV risk stratification or as an opportunistic screening tool requires prospective evaluation.

Alberto Presta, Enzo Tartaglione, Attilio Fiandrotti et al.

Learned progressive image compression is gaining momentum as it allows improved image reconstruction as more bits are decoded at the receiver. We propose a progressive image compression method in which an image is first represented as a pair of base-quality and top-quality latent representations. Next, a residual latent representation is encoded as the element-wise difference between the top and base representations. Our scheme enables progressive image compression with element-wise granularity by introducing a masking system that ranks each element of the residual latent representation from most to least important, dividing it into complementary components, which can be transmitted separately to the decoder in order to obtain different reconstruction quality. The masking system does not add further parameters nor complexity. At the receiver, any elements of the top latent representation excluded from the transmitted components can be independently replaced with the mean predicted by the hyperprior architecture, ensuring reliable reconstructions at any intermediate quality level. We also introduced Rate Enhancement Modules (REMs), which refine the estimation of entropy parameters using already decoded components. We obtain results competitive with state-of-the-art competitors, while significantly reducing computational complexity, decoding time, and number of parameters.