We introduce LOKI, a compute-efficient framework for co-designing morphologies and control policies that generalize across unseen tasks. Inspired by biological adaptation—where animals quickly adjust to morphological changes—our method overcomes the inefficiencies of traditional evolutionary and quality-diversity algorithms. We propose learning convergent functions: shared control policies trained across clusters of morphologically similar designs in a learned latent space, drastically reducing the training cost per design. Simultaneously, we promote divergent forms by replacing mutation with dynamic local search, enabling broader exploration and preventing premature convergence. The policy reuse allows us to explore $\sim780\times$ more designs using 78\% fewer simulation steps and 40\% less compute per design. Local competition paired with a broader search results in a diverse set of high-performing final morphologies. Using the UNIMAL design space and a flat-terrain locomotion task, LOKI discovers a rich variety of designs—ranging from quadrupeds to crabs, bipedals, and spinners—far more diverse than those produced by prior work. These morphologies also transfer better to unseen downstream tasks in agility, stability, and manipulation domains (e.g. $2 \times$ higher reward on bump and push box incline tasks). Overall, our approach produces designs that are both diverse and adaptable, with substantially greater sample efficiency than existing co-design methods.

Planning with pretrained diffusion models has emerged as a promising approach for solving test-time guided control problems. Standard gradient guidance typically performs optimally under convex, differentiable reward landscapes. However, it shows substantially reduced effectiveness in real-world scenarios with non-convex objectives, non-differentiable constraints, and multi-reward structures. Furthermore, recent supervised planning approaches require task-specific training or value estimators, which limits test-time flexibility and zero-shot generalization. We propose a Tree-guided Diffusion Planner (TDP), a zero-shot test-time planning framework that balances exploration and exploitation through structured trajectory generation. We frame test-time planning as a tree search problem using a bi-level sampling process: (1) diverse parent trajectories are produced via training-free particle guidance to encourage broad exploration, and (2) sub-trajectories are refined through fast conditional denoising guided by task objectives. TDP addresses the limitations of gradient guidance by exploring diverse trajectory regions and harnessing gradient information across this expanded solution space using only pretrained models and test-time reward signals. We evaluate TDP on three diverse tasks: maze gold-picking, robot arm block manipulation, and AntMaze multi-goal exploration. TDP consistently outperforms state-of-the-art approaches on all tasks. The project page can be found at: https://tree-diffusion-planner.github.io

In the present era of deep learning, continual learning research is mainly focused on mitigating forgetting when training a neural network with stochastic gradient descent on a non-stationary stream of data. On the other hand, in the more classical literature of statistical machine learning, many models have sequential Bayesian update rules that yield the same learning outcome as the batch training, i.e., they are completely immune to catastrophic forgetting. However, they are often overly simple to model complex real-world data. In this work, we adopt the meta-learning paradigm to combine the strong representational power of neural networks and simple statistical models’ robustness to forgetting. In our novel meta-continual learning framework, continual learning takes place only in statistical models via ideal sequential Bayesian update rules, while neural networks are meta-learned to bridge the raw data and the statistical models. Since the neural networks remain fixed during continual learning, they are protected from catastrophic forgetting. This approach not only achieves significantly improved performance but also exhibits excellent scalability. Since our approach is domain-agnostic and model-agnostic, it can be applied to a wide range of problems and easily integrated with existing model architectures.

Generally, object detection on Synthetic-Aperture Radar (SAR) images is known to be more challenging than that in Electro-Optical (EO) satellite images because SAR images have non-negligible speckle noise and require extensive data pre-processing. Nevertheless, object detection in SAR images is important, as SAR imagery can be obtained under severe weather and time conditions. While many recent object detection approaches on SAR imagery focus on improving detection accuracy, few studies focus on improving processing efficiency. In fact, there are significant challenges and trade-offs to achieve both high accuracy and efficiency at the same time. In this work, we introduce SAROD, a novel efficient end-to-end object detection framework on SAR images based on Reinforcement Learning (RL) to balance the tradeoffs. Our proposed model consists of two detectors, coarse and fine-grained detectors, with an RL agent, where RL has not yet been utilized for object detection on SAR imagery. Our model was evaluated on a challenging SAR imagery dataset, achieving performance comparable to state-of-the-art detectors while maintaining high efficiency of source data usage.

Recent advancements in Generative Adversarial Networks (GANs) enable the generation of highly realistic images, raising concerns about their misuse for malicious purposes. Detecting these GAN-generated images (GAN-images) becomes increasingly challenging due to the significant reduction of underlying artifacts and specific patterns. The absence of such traces can hinder detection algorithms from identifying GAN-images and transferring knowledge to identify other types of GAN-images as well. In this work, we present the Transferable GAN-images Detection framework T-GD, a robust transferable framework for an effective detection of GAN-images. T-GD is composed of a teacher and a student model that can iteratively teach and evaluate each other to improve the detection performance. First, we train the teacher model on the source dataset and use it as a starting point for learning the target dataset. To train the student model, we inject noise by mixing up the source and target datasets, while constraining the weight variation to preserve the starting point. Our approach is a self-training method, but distinguishes itself from prior approaches by focusing on improving the transferability of GAN-image detection. T-GD achieves high performance on the source dataset by overcoming catastrophic forgetting and effectively detecting state-of-the-art GAN-images with only a small volume of data without any metadata information.

Proteins interact with each other to regulate their functionality and localization. The accumulated protein interaction evidences are represented by protein interaction network using a graph abstraction. Topological properties of protein interaction networks have been explored to characterize proteins and predict undiscovered interactions. Meanwhile, many researchers have tried to explain how protein interaction network is formed through evolutionary process. While one group of researchers have made efforts to simulate current protein interaction starting from hypothetical infant state of protein interaction networks through suggested evolutionary models, another group of researchers have made efforts to estimate the phylogenetic age of proteins from evolutionary relationship. Recently, these efforts gave rise to the database of phylogenetic age of proteins and this allows many researchers to estimate phylogenetic age of proteins of their interest easily. As seen by their terms, the evolutionary model of protein interaction networks and phylogenetic age of proteins are closely related, thus topological properties of protein interactions, which is important in studies of the evolutionary models, can be linked to the phylogenetic age of proteins. In this paper, we construct a weighted human protein interaction network from a human protein interaction network, which is provided by BioGRID database. The weight of an edge is defined as the number of triangles which contains this edge in the protein interaction network and we call this weight as the triangle score. From the weighted protein interaction network, we extract proteins that are incident to an edge that has a high triangle score. We obtain phylogenetic age of proteins and measure various statistical values to observe correlation between phylogenetic age and the triangle score. As a result, we show that the proteins, that participate in interactions with high triangle score, are old in terms of phylogenetic age. Also we show that for interactions within a same phylogenetic age category tend to have higher triangle scores than the interactions with only one of its participant protein contained in the given age category.