Low-rank adaptation (LoRA) has been demonstrated effective in reducing the trainable parameter number when fine-tuning a large foundation model (LLM). However, it still encounters computational and memory challenges when scaling to larger models or addressing more complex task adaptation. In this work, we introduce Sparse Spectrum Adaptation via Discrete Hartley Transformation (SSH), a novel approach that significantly reduces the number of trainable parameters while enhancing model performance. It selects the most informative spectral components across all layers, under the guidance of the initial weights after a discrete Hartley transformation (DHT). The lightweight inverse DHT then projects the spectrum back into the spatial domain for updates. Extensive experiments across both single-modality tasks such as language understanding and generation and multi-modality tasks such as video-text understanding demonstrate that SSH outperforms existing parameter-efficient fine-tuning (PEFT) methods while achieving substantial reductions in computational cost and memory requirements.

Satellite altimetry has been widely utilized to monitor global sea surface dynamics, enabling investigation of upper ocean variability from basin-scale to localized eddy ranges. However, the sparse spatial resolution of observational altimetry limits our understanding of oceanic submesoscale variability, prevalent at horizontal scales below 0.25o resolution. Here, we introduce a state-of-the-art generative diffusion model to train high-resolution sea surface height (SSH) reanalysis data and demonstrate its advantage in observational SSH downscaling over the eddy-rich Kuroshio Extension region. The diffusion-based model effectively downscales raw satellite-interpolated data from 0.25o resolution to 1/16o, corresponding to approximately 12-km wavelength. This model outperforms other high-resolution reanalysis datasets and neural network-based methods. Also, it successfully reproduces the spatial patterns and power spectra of satellite along-track observations. Our diffusion-based results indicate that eddy kinetic energy at horizontal scales less than 250 km has intensified significantly since 2004 in the Kuroshio Extension region. These findings underscore the great potential of deep learning in reconstructing satellite altimetry and enhancing our understanding of ocean dynamics at eddy scales.

Now we can do our favorite two things and update our packages and repositories. Something to note is that the package manager of course will depend on the distribution you chose. For RedHat it can be either dnf or yum, Debian(or Ubuntu) will use apt, Arch will use Pacman, and openSuse will use man. So if you didn't choose to use RedHat, just replace my dnf with your respective package manager. After pressing y and enter at least once, you are now going have to get your new best friend: SSH.

Sea surface height (SSH) is a key geophysical parameter for monitoring and studying meso-scale surface ocean dynamics. For several decades, the mapping of SSH products at regional and global scales has relied on nadir satellite altimeters, which provide one-dimensional-only along-track satellite observations of the SSH. The Surface Water and Ocean Topography (SWOT) mission deploys a new sensor that acquires for the first time wide-swath two-dimensional observations of the SSH. This provides new means to observe the ocean at previously unresolved spatial scales. A critical challenge for the exploiting of SWOT data is the separation of the SSH from other signals present in the observations. In this paper, we propose a novel learning-based approach for this SWOT calibration problem. It benefits from calibrated nadir altimetry products and a scale-space decomposition adapted to SWOT swath geometry and the structure of the different processes in play. In a supervised setting, our method reaches the state-of-the-art residual error of ~1.4cm while proposing a correction on the entire spectral from 10km to 1000k

In this work we improve forecasting of Sea Surface A recent and promising line of work consists of combining Height (SSH) and current velocity (speed and direction) ML with physics-based models -- often referred to as in oceanic scenarios. We do so by resorting Physics-Informed Machine Learning (PIML). Such an approach to Random Forests so as to predict the error of a numerical aims to take advantage of both the power of pattern forecasting system developed for the Santos recognition given by ML approaches and the power of generalization Channel in Brazil. We have used the Santos Operational in unseen scenarios given by the physics-based Forecasting System (SOFS) and data collected model. in situ between the years of 2019 and 2021. This work expands on our previous work [Moreno et al., In previous studies we have applied similar methods 2022] where PIML was used to correct the error predicted for current velocity in the channel entrance, in by a numerical model of the speed of water current in a this work we expand the application to improve the measuring station. Our main contribution here consists of SHH forecast and include four other stations in the inserting a correction for the direction of the water current channel. We have obtained an average reduction and the sea surface height (SSH) predicted by the numerical of 11.9% in forecasting Root-Mean Square Error model into the PIML model. In addition, we expand the (RMSE) and 38.7% in bias with our approach. We corrections to other measurement stations in the Santos-São also obtained an increase of Agreement (IOA) in 10 Vicente-Bertioga Estuarine System region on the Brazilian of the 14 combinations of forecasted variables and coast.

In this paper, at first, the impact of ImageNet pre-training on Facial Expression Recognition (FER) was tested under different augmentation levels. It could be seen from the results that training from scratch could reach better performance compared to ImageNet fine-tuning at stronger augmentation levels. After that, a framework was proposed for standard Supervised Learning (SL), called Hybrid Learning (HL) which used Self-Supervised co-training with SL in Multi-Task Learning (MTL) manner. Leveraging Self-Supervised Learning (SSL) could gain additional information from input data like spatial information from faces which helped the main SL task. It is been investigated how this method could be used for FER problems with self-supervised pre-tasks such as Jigsaw puzzling and in-painting. The supervised head (SH) was helped by these two methods to lower the error rate under different augmentations and low data regime in the same training settings. The state-of-the-art was reached on AffectNet via two completely different HL methods, without utilizing additional datasets. Moreover, HL's effect was shown on two different facial-related problem, head poses estimation and gender recognition, which concluded to reduce in error rate by up to 9% and 1% respectively. Also, we saw that the HL methods prevented the model from reaching overfitting.

SSH is obviously encrypted, yet valuable contextual information still exists within the network traffic that can go towards TTP's, intent, success and magnitude of actions on objectives. There may even exist situations where valuable context is not available or deleted from hosts, and so having an immutable and un-alterable passive network capture gives additional forensic context. Separately to the forensic context, packet strider predictions could also be used in an active fashion, for example to shun/RST forward connections if a tunneled reverse SSH session initiation feature is predicted within, even before reverse authentication is offered. The pcap "forward_reverse.pcap" is from a common TTP of a Reverse SSH shell, a favorite of red teams everywhere. Now on the attacker's machine (the server), a reverse shell is initiated back to the victim: Then finally with the Forward session the session is closed, just to demonstrate that the forward SSH feature detection still works.