Slicing Vision Transformer for Flexible Inference

Vision Transformers (ViT) is known for its scalability. In this work, we target to scale down a ViT to fit in an environment with dynamic-changing resource constraints. We observe that smaller ViTs are intrinsically the sub-networks of a larger ViT with different widths. Thus, we propose a general framework, named Scala, to enable a single network to represent multiple smaller ViTs with flexible inference capability, which aligns with the inherent design of ViT to vary from widths. Concretely, Scala activates several subnets during training, introduces Isolated Activation to disentangle the smallest sub-network from other subnets, and leverages Scale Coordination to ensure each sub-network receives simplified, steady, and accurate learning objectives. Comprehensive empirical validations on different tasks demonstrate that with only one-shot training, Scala learns slimmable representation without modifying the original ViT structure and matches the performance of Separate Training. Compared with the prior art, Scala achieves an average improvement of 1.6% on ImageNet-1K with fewer parameters.

Foundation Model for Composite Materials and Microstructural Analysis

The rapid advancement of machine learning has unlocked numerous opportunities for materials science, particularly in accelerating the design and analysis of materials. However, a significant challenge lies in the scarcity and high cost of obtaining high-quality materials datasets. In other fields, such as natural language processing, foundation models pre-trained on large datasets have achieved exceptional success in transfer learning, effectively leveraging latent features to achieve high performance on tasks with limited data. Despite this progress, the concept of foundation models remains underexplored in materials science. Here, we present a foundation model specifically designed for composite materials. Our model is pre-trained on a dataset of short-fiber composites to learn robust latent features. During transfer learning, the MMAE accurately predicts homogenized stiffness, with an R2 score reaching as high as 0.959 and consistently exceeding 0.91, even when trained on limited data. These findings validate the feasibility and effectiveness of foundation models in composite materials. We anticipate extending this approach to more complex three-dimensional composite materials, polycrystalline materials, and beyond. Moreover, this framework enables high-accuracy predictions even when experimental data are scarce, paving the way for more efficient and cost-effective materials design and analysis.

Radio-Based Passive Target Tracking by a Mobile Receiver with Unknown Transmitter Position

In this paper, we propose a radio-based passive target tracking algorithm using multipath measurements, including the angle of arrival and relative distance. We focus on a scenario in which a mobile receiver continuously receives radio signals from a transmitter located at an unknown position. The receiver utilizes multipath measurements extracted from the received signal to jointly localize the transmitter and the scatterers over time, with scatterers comprising a moving target and stationary objects that can reflect signals within the environment. We develop a comprehensive probabilistic model for the target tracking problem, incorporating the localization of the transmitter and scatterers, the identification of false alarms and missed detections in the measurements, and the association between scatterers and measurements. We employ a belief propagation approach to compute the posterior distributions of the positions of the scatterers and the transmitter. Additionally, we introduce a particle implementation for the belief propagation method. Simulation results demonstrate that our proposed algorithm outperforms existing benchmark methods in terms of target tracking accuracy.

Age of Information-Oriented Probabilistic Link Scheduling for Device-to-Device Networks

This paper focuses on optimizing the long-term average age of information (AoI) in device-to-device (D2D) networks through age-aware link scheduling. The problem is naturally formulated as a Markov decision process (MDP). However, finding the optimal policy for the formulated MDP in its original form is challenging due to the intertwined AoI dynamics of all D2D links. To address this, we propose an age-aware stationary randomized policy that determines the probability of scheduling each link in each time slot based on the AoI of all links and the statistical channel state information among all transceivers. By employing the Lyapunov optimization framework, our policy aims to minimize the Lyapunov drift in every time slot. Nonetheless, this per-slot minimization problem is nonconvex due to cross-link interference in D2D networks, posing significant challenges for real-time decision-making. After analyzing the permutation equivariance property of the optimal solutions to the per-slot problem, we apply a message passing neural network (MPNN), a type of graph neural network that also exhibits permutation equivariance, to optimize the per-slot problem in an unsupervised learning manner. Simulation results demonstrate the superior performance of the proposed age-aware stationary randomized policy over baselines and validate the scalability of our method.

Robust Guided Diffusion for Offline Black-Box Optimization

Offline black-box optimization aims to maximize a black-box function using an offline dataset of designs and their measured properties. Two main approaches have emerged: the forward approach, which learns a mapping from input to its value, thereby acting as a proxy to guide optimization, and the inverse approach, which learns a mapping from value to input for conditional generation. (a) Although proxy-free~(classifier-free) diffusion shows promise in robustly modeling the inverse mapping, it lacks explicit guidance from proxies, essential for generating high-performance samples beyond the training distribution. Therefore, we propose \textit{proxy-enhanced sampling} which utilizes the explicit guidance from a trained proxy to bolster proxy-free diffusion with enhanced sampling control. (b) Yet, the trained proxy is susceptible to out-of-distribution issues. To address this, we devise the module \textit{diffusion-based proxy refinement}, which seamlessly integrates insights from proxy-free diffusion back into the proxy for refinement. To sum up, we propose \textit{\textbf{R}obust \textbf{G}uided \textbf{D}iffusion for Offline Black-box Optimization}~(\textbf{RGD}), combining the advantages of proxy~(explicit guidance) and proxy-free diffusion~(robustness) for effective conditional generation. RGD achieves state-of-the-art results on various design-bench tasks, underscoring its efficacy. Our code is at https://anonymous.4open.science/r/RGD-27A5/README.md.

Bilingual Adaptation of Monolingual Foundation Models

We present an efficient method for adapting a monolingual Large Language Model (LLM) to another language, addressing challenges of catastrophic forgetting and tokenizer limitations. We focus this study on adapting Llama 2 to Arabic. Our two-stage approach begins with expanding the vocabulary and training only the embeddings matrix, followed by full model continual pretraining on a bilingual corpus. By continually pretraining on a mix of Arabic and English corpora, the model retains its proficiency in English while acquiring capabilities in Arabic. Our approach results in significant improvements in Arabic and slight enhancements in English, demonstrating cost-effective cross-lingual transfer. We also perform extensive ablations on embedding initialization techniques, data mix ratios, and learning rates and release a detailed training recipe.

Physics-Informed Geometric Operators to Support Surrogate, Dimension Reduction and Generative Models for Engineering Design

In this work, we propose a set of physics-informed geometric operators (GOs) to enrich the geometric data provided for training surrogate/discriminative models, dimension reduction, and generative models, typically employed for performance prediction, dimension reduction, and creating data-driven parameterisations, respectively. However, as both the input and output streams of these models consist of low-level shape representations, they often fail to capture shape characteristics essential for performance analyses. Therefore, the proposed GOs exploit the differential and integral properties of shapes--accessed through Fourier descriptors, curvature integrals, geometric moments, and their invariants--to infuse high-level intrinsic geometric information and physics into the feature vector used for training, even when employing simple model architectures or low-level parametric descriptions. We showed that for surrogate modelling, along with the inclusion of the notion of physics, GOs enact regularisation to reduce over-fitting and enhance generalisation to new, unseen designs. Furthermore, through extensive experimentation, we demonstrate that for dimension reduction and generative models, incorporating the proposed GOs enriches the training data with compact global and local geometric features. This significantly enhances the quality of the resulting latent space, thereby facilitating the generation of valid and diverse designs. Lastly, we also show that GOs can enable learning parametric sensitivities to a great extent. Consequently, these enhancements accelerate the convergence rate of shape optimisers towards optimal solutions.

Hardware-Aware Parallel Prompt Decoding for Memory-Efficient Acceleration of LLM Inference

The auto-regressive decoding of Large Language Models (LLMs) results in significant overheads in their hardware performance. While recent research has investigated various speculative decoding techniques for multi-token generation, these efforts have primarily focused on improving processing speed such as throughput. Crucially, they often neglect other metrics essential for real-life deployments, such as memory consumption and training cost. To overcome these limitations, we propose a novel parallel prompt decoding that requires only $0.0002$% trainable parameters, enabling efficient training on a single A100-40GB GPU in just 16 hours. Inspired by the human natural language generation process, $PPD$ approximates outputs generated at future timesteps in parallel by using multiple prompt tokens. This approach partially recovers the missing conditional dependency information necessary for multi-token generation, resulting in up to a 28% higher acceptance rate for long-range predictions. Furthermore, we present a hardware-aware dynamic sparse tree technique that adaptively optimizes this decoding scheme to fully leverage the computational capacities on different GPUs. Through extensive experiments across LLMs ranging from MobileLlama to Vicuna-13B on a wide range of benchmarks, our approach demonstrates up to 2.49$\times$ speedup and maintains a minimal runtime memory overhead of just $0.0004$%. More importantly, our parallel prompt decoding can serve as an orthogonal optimization for synergistic integration with existing speculative decoding, showing up to $1.22\times$ further speed improvement. Our code is available at https://github.com/hmarkc/parallel-prompt-decoding.

Pearls from Pebbles: Improved Confidence Functions for Auto-labeling

Auto-labeling is an important family of techniques that produce labeled training sets with minimum manual labeling. A prominent variant, threshold-based auto-labeling (TBAL), works by finding a threshold on a model's confidence scores above which it can accurately label unlabeled data points. However, many models are known to produce overconfident scores, leading to poor TBAL performance. While a natural idea is to apply off-the-shelf calibration methods to alleviate the overconfidence issue, such methods still fall short. Rather than experimenting with ad-hoc choices of confidence functions, we propose a framework for studying the \emph{optimal} TBAL confidence function. We develop a tractable version of the framework to obtain \texttt{Colander} (Confidence functions for Efficient and Reliable Auto-labeling), a new post-hoc method specifically designed to maximize performance in TBAL systems. We perform an extensive empirical evaluation of our method \texttt{Colander} and compare it against methods designed for calibration. \texttt{Colander} achieves up to 60\% improvements on coverage over the baselines while maintaining auto-labeling error below $5\%$ and using the same amount of labeled data as the baselines.

3D object quality prediction for Metal Jet Printer with Multimodal thermal encoder

With the advancements in 3D printing technologies, it is extremely important that the quality of 3D printed objects, and dimensional accuracies should meet the customer's specifications. Various factors during metal printing affect the printed parts' quality, including the power quality, the printing stage parameters, the print part's location inside the print bed, the curing stage parameters, and the metal sintering process. With the large data gathered from HP's MetJet printing process, AI techniques can be used to analyze, learn, and effectively infer the printed part quality metrics, as well as assist in improving the print yield. In-situ thermal sensing data captured by printer-installed thermal sensors contains the part thermal signature of fusing layers. Such part thermal signature contains a convoluted impact from various factors. In this paper, we use a multimodal thermal encoder network to fuse data of a different nature including the video data vectorized printer control data, and exact part thermal signatures with a trained encoder-decoder module. We explored the data fusing techniques and stages for data fusing, the optimized end-to-end model architecture indicates an improved part quality prediction accuracy.