Search

We present the Differentially Private Blockchain-Based Vertical Federal Learning (DP-BBVFL) algorithm that provides verifiability and privacy guarantees for decentralized applications. DP-BBVFL uses a smart contract to aggregate the feature representations, i.e., the embeddings, from clients transparently. We apply local differential privacy to provide privacy for embeddings stored on a blockchain, hence protecting the original data. We provide the first prototype application of differential privacy with blockchain for vertical federated learning. Our experiments with medical data show that DP-BBVFL achieves high accuracy with a tradeoff in training time due to on-chain aggregation. This innovative fusion of differential privacy and blockchain technology in DP-BBVFL could herald a new era of collaborative and trustworthy machine learning applications across several decentralized application domains.

This paper characterizes the microdosimetric spectra of a single-energy carbon-ion pencil beam at MedAustron using a miniature solid-state silicon microdosimeter to estimate the impact of the lateral distribution of the different fragments on the microdosimetric spectra. The microdosimeter was fixed at one depth and then laterally moved away from the central beam axis in steps of approximately 2 mm. The measurements were taken in both horizontal and vertical direction in a water phantom at different depths. In a position on the distal dose fall-off beyond the Bragg peak, the frequency-mean and the dose-mean lineal energies were derived using either the entire range of y-values, or a sub-range of y values, presumingly corresponding mainly to contributions from primary particles. The measured microdosimetric spectra do not exhibit a significant change up to 4 mm away from the beam central axis. For lateral positions more than 4 mm away from the central axis, the relative contribution of the lower lineal-energy part of the spectrum increases with lateral distance due to the increased partial dose from secondary fragments. The average values yF and yD are almost constant for each partial contribution. However, when all particles are considered together, the average value of yF and yD varies with distance from the axis due to the changing dose fractions of these two components varying by 30 % and 10 % respectively up to the most off axis vertical position. Characteristic features in the microdosimetric spectra providing strong indications of the presence of helium and boron fragments have been observed downstream of the distal part of the Bragg peak. We were able to investigate the radiation quality as function of off-axis position. These measurements emphasize variation of the radiation quality within the beam and this has implications in terms of relative biological effectiveness., Comment: 19 pages, 9 figures

Video motion magnification is a technique to capture and amplify subtle motion in a video that is invisible to the naked eye. The deep learning-based prior work successfully demonstrates the modelling of the motion magnification problem with outstanding quality compared to conventional signal processing-based ones. However, it still lags behind real-time performance, which prevents it from being extended to various online applications. In this paper, we investigate an efficient deep learning-based motion magnification model that runs in real time for full-HD resolution videos. Due to the specified network design of the prior art, i.e. inhomogeneous architecture, the direct application of existing neural architecture search methods is complicated. Instead of automatic search, we carefully investigate the architecture module by module for its role and importance in the motion magnification task. Two key findings are 1) Reducing the spatial resolution of the latent motion representation in the decoder provides a good trade-off between computational efficiency and task quality, and 2) surprisingly, only a single linear layer and a single branch in the encoder are sufficient for the motion magnification task. Based on these findings, we introduce a real-time deep learning-based motion magnification model with4.2X fewer FLOPs and is 2.7X faster than the prior art while maintaining comparable quality., Comment: 19 pages

Prescriptive AI represents a transformative shift in decision-making, offering causal insights and actionable recommendations. Despite its huge potential, enterprise adoption often faces several challenges. The first challenge is caused by the limitations of observational data for accurate causal inference which is typically a prerequisite for good decision-making. The second pertains to the interpretability of recommendations, which is crucial for enterprise decision-making settings. The third challenge is the silos between data scientists and business users, hindering effective collaboration. This paper outlines an initiative from IBM Research, aiming to address some of these challenges by offering a suite of prescriptive AI solutions. Leveraging insights from various research papers, the solution suite includes scalable causal inference methods, interpretable decision-making approaches, and the integration of large language models (LLMs) to bridge communication gaps via a conversation agent. A proof-of-concept, PresAIse, demonstrates the solutions' potential by enabling non-ML experts to interact with prescriptive AI models via a natural language interface, democratizing advanced analytics for strategic decision-making., Comment: 14 pages

Binary neural networks (BNNs) have been widely adopted to reduce the computational cost and memory storage on edge-computing devices by using one-bit representation for activations and weights. However, as neural networks become wider/deeper to improve accuracy and meet practical requirements, the computational burden remains a significant challenge even on the binary version. To address these issues, this paper proposes a novel method called Minimum Spanning Tree (MST) compression that learns to compress and accelerate BNNs. The proposed architecture leverages an observation from previous works that an output channel in a binary convolution can be computed using another output channel and XNOR operations with weights that differ from the weights of the reused channel. We first construct a fully connected graph with vertices corresponding to output channels, where the distance between two vertices is the number of different values between the weight sets used for these outputs. Then, the MST of the graph with the minimum depth is proposed to reorder output calculations, aiming to reduce computational cost and latency. Moreover, we propose a new learning algorithm to reduce the total MST distance during training. Experimental results on benchmark models demonstrate that our method achieves significant compression ratios with negligible accuracy drops, making it a promising approach for resource-constrained edge-computing devices., Comment: 11 pages, 9 figures, ICCV 2023

6-DoF pose estimation is an essential component of robotic manipulation pipelines. However, it usually suffers from a lack of generalization to new instances and object types. Most widely used methods learn to infer the object pose in a discriminative setup where the model filters useful information to infer the exact pose of the object. While such methods offer accurate poses, the model does not store enough information to generalize to new objects. In this work, we address the generalization capability of pose estimation using models that contain enough information about the object to render it in different poses. We follow the line of work that inverts neural renderers to infer the pose. We propose i-$\sigma$SRN to maximize the information flowing from the input pose to the rendered scene and invert them to infer the pose given an input image. Specifically, we extend Scene Representation Networks (SRNs) by incorporating a separate network for density estimation and introduce a new way of obtaining a weighted scene representation. We investigate several ways of initial pose estimates and losses for the neural renderer. Our final evaluation shows a significant improvement in inference performance and speed compared to existing approaches.

The "majority dynamics" process on a social network begins with an initial phase, where the individuals are split into two competing parties, Red and Blue. Every day, everyone updates their affiliation to match the majority among those of their friends. While studying this process on Erdos-Renyi G(n, p) random graph (with constant density), the authors discovered the "Power of Few" phenomenon, showing that a very small advantage to one side already guarantees that everybody will unanimously join that side after just a few days with overwhelming probability. For example, when p = 1/2, then 10 extra members guarantee this unanimity with a 90% chance, regardless of the value of n. In this paper, we study this phenomenon for sparse random graphs. It is clear that below the connectivity threshold, the phenomenon ceases to hold, as the isolated vertices never change their colors. We show that it holds for every density above the threshold. To make the process more realistic, we also assume that individuals can randomly activate their accounts to post their opinions and observe their neighbors (just as we do on social media). We prove that the phenomenon is robust under this assumption., Comment: 27 pages

A fundamental challenge of over-parameterized deep learning models is learning meaningful data representations that yield good performance on a downstream task without over-fitting spurious input features. This work proposes MaskTune, a masking strategy that prevents over-reliance on spurious (or a limited number of) features. MaskTune forces the trained model to explore new features during a single epoch finetuning by masking previously discovered features. MaskTune, unlike earlier approaches for mitigating shortcut learning, does not require any supervision, such as annotating spurious features or labels for subgroup samples in a dataset. Our empirical results on biased MNIST, CelebA, Waterbirds, and ImagenNet-9L datasets show that MaskTune is effective on tasks that often suffer from the existence of spurious correlations. Finally, we show that MaskTune outperforms or achieves similar performance to the competing methods when applied to the selective classification (classification with rejection option) task. Code for MaskTune is available at https://github.com/aliasgharkhani/Masktune., Comment: Accepted to NeurIPS 2022

In this note, we verify that the bounded shadowing property and quasi-hyperbolicity of bounded linear operators on Hilbert spaces are preserved under Aluthge transforms.

In this note, we introduce the notion of $r$-homoclinic points. We show that an operator on a Banach space is hyperbolic if and only if it is shadowing and has no nonzero $r$-homoclinic points. We also solve invariant subspace problem (ISP for brevity) for shadowing operators on Banach spaces. Afterwards, we verify that the set of generalized hyperbolic operators is invariant under $\lambda$-Aluthge transforms for every $\lambda \in \left( 0,1 \right)$. Next, the Aluthge iterates of invertible operators converge to hyperbolic operators only if the initial operators are hyperbolic. Finally, we prove that the Aluthge iterates of shifted hyperbolic bilateral weighted shifts diverge and that hyperbolic bilateral weighted shifts with divergent Aluthge iterates exist.

This paper constitutes the novel hypergraph convolutional neural networkbased clustering technique. This technique is employed to solve the clustering problem for the Citeseer dataset and the Cora dataset. Each dataset contains the feature matrix and the incidence matrix of the hypergraph (i.e., constructed from the feature matrix). This novel clustering method utilizes both matrices. Initially, the hypergraph auto-encoders are employed to transform both the incidence matrix and the feature matrix from high dimensional space to low dimensional space. In the end, we apply the k-means clustering technique to the transformed matrix. The hypergraph convolutional neural network (CNN)-based clustering technique presented a better result on performance during experiments than those of the other classical clustering techniques.

User modeling is crucial to understanding user behavior and essential for improving user experience and personalized recommendations. When users interact with software, vast amounts of command sequences are generated through logging and analytics systems. These command sequences contain clues to the users' goals and intents. However, these data modalities are highly unstructured and unlabeled, making it difficult for standard predictive systems to learn from. We propose SimCURL, a simple yet effective contrastive self-supervised deep learning framework that learns user representation from unlabeled command sequences. Our method introduces a user-session network architecture, as well as session dropout as a novel way of data augmentation. We train and evaluate our method on a real-world command sequence dataset of more than half a billion commands. Our method shows significant improvement over existing methods when the learned representation is transferred to downstream tasks such as experience and expertise classification.

Batch normalization (BN) is a ubiquitous technique for training deep neural networks that accelerates their convergence to reach higher accuracy. However, we demonstrate that BN comes with a fundamental drawback: it incentivizes the model to rely on low-variance features that are highly specific to the training (in-domain) data, hurting generalization performance on out-of-domain examples. In this work, we investigate this phenomenon by first showing that removing BN layers across a wide range of architectures leads to lower out-of-domain and corruption errors at the cost of higher in-domain errors. We then propose Counterbalancing Teacher (CT), a method which leverages a frozen copy of the same model without BN as a teacher to enforce the student network's learning of robust representations by substantially adapting its weights through a consistency loss function. This regularization signal helps CT perform well in unforeseen data shifts, even without information from the target domain as in prior works. We theoretically show in an overparameterized linear regression setting why normalization leads to a model's reliance on such in-domain features, and empirically demonstrate the efficacy of CT by outperforming several baselines on robustness benchmarks such as CIFAR-10-C, CIFAR-100-C, and VLCS.

Background: Access to medical care is strongly dependent on resource allocation, such as the geographical distribution of medical facilities. Nevertheless, this data is usually restricted to country official documentation, not available to the public. While some medical facilities' data is accessible as semantic resources on the Web, it is not consistent in its modeling and has yet to be integrated into a complete, open, and specialized repository. This work focuses on generating a comprehensive semantic dataset of medical facilities worldwide containing extensive information about such facilities' geo-location. Results: For this purpose, we collect, align, and link various open-source databases where medical facilities' information may be present. This work allows us to evaluate each data source along various dimensions, such as completeness, correctness, and interlinking with other sources, all critical aspects of current knowledge representation technologies. Conclusions: Our contributions directly benefit stakeholders in the biomedical and health domain (patients, healthcare professionals, companies, regulatory authorities, and researchers), who will now have a better overview of the access to and distribution of medical facilities., Comment: Accepted at 4th Workshop on Semantic Web solutions for large-scale biomedical data analytics (SeWeBMeDA-2020)

Constrained optimization problems can be difficult because their search spaces have properties not conducive to search, e.g., multimodality, discontinuities, or deception. To address such difficulties, considerable research has been performed on creating novel evolutionary algorithms or specialized genetic operators. However, if the representation that defined the search space could be altered such that it only permitted valid solutions that satisfied the constraints, the task of finding the optimal would be made more feasible without any need for specialized optimization algorithms. We propose Constrained Optimization in Latent Space (COIL), which uses a VAE to generate a learned latent representation from a dataset comprising samples from the valid region of the search space according to a constraint, thus enabling the optimizer to find the objective in the new space defined by the learned representation. Preliminary experiments show promise: compared to an identical GA using a standard representation that cannot meet the constraints or find fit solutions, COIL with its learned latent representation can perfectly satisfy different types of constraints while finding high-fitness solutions., Comment: Genetic and Evolutionary Computation Conference Companion (GECCO '22 Companion). ACM, Boston, USA, 8 pages

Physical products are often complex assemblies combining a multitude of 3D parts modeled in computer-aided design (CAD) software. CAD designers build up these assemblies by aligning individual parts to one another using constraints called joints. In this paper we introduce JoinABLe, a learning-based method that assembles parts together to form joints. JoinABLe uses the weak supervision available in standard parametric CAD files without the help of object class labels or human guidance. Our results show that by making network predictions over a graph representation of solid models we can outperform multiple baseline methods with an accuracy (79.53%) that approaches human performance (80%). Finally, to support future research we release the Fusion 360 Gallery assembly dataset, containing assemblies with rich information on joints, contact surfaces, holes, and the underlying assembly graph structure., Comment: CVPR 2022; code available at https://github.com/AutodeskAILab/JoinABLe and data available at https://github.com/AutodeskAILab/Fusion360GalleryDataset

Learning interpretable and human-controllable representations that uncover factors of variation in data remains an ongoing key challenge in representation learning. We investigate learning group-disentangled representations for groups of factors with weak supervision. Existing techniques to address this challenge merely constrain the approximate posterior by averaging over observations of a shared group. As a result, observations with a common set of variations are encoded to distinct latent representations, reducing their capacity to disentangle and generalize to downstream tasks. In contrast to previous works, we propose GroupVAE, a simple yet effective Kullback-Leibler (KL) divergence-based regularization across shared latent representations to enforce consistent and disentangled representations. We conduct a thorough evaluation and demonstrate that our GroupVAE significantly improves group disentanglement. Further, we demonstrate that learning group-disentangled representations improve upon downstream tasks, including fair classification and 3D shape-related tasks such as reconstruction, classification, and transfer learning, and is competitive to supervised methods.