기법

DetNAS is a neural architecture search algorithm for the design of better backbones for object detection. It is based on the technique of one-shot supernet, which contains all possible networks in the search space. The supernet is trained under the typical detector training schedule: ImageNet pre-training and detection fine-tuning. Then, the architecture search is performed on the trained supernet, using the detection task as the guidance. DetNAS uses evolutionary search as opposed to RL-based methods or gradient-based methods.

Differentiable Augmentation (DiffAugment) is a set of differentiable image transformations used to augment data during GAN training. The transformations are applied to the real and generated images. It enables the gradients to be propagated through the augmentation back to the generator, regularizes the discriminator without manipulating the target distribution, and maintains the balance of training dynamics. Three choices of transformation are preferred by the authors in their experiments: Translation, CutOut, and Color.

Differential Diffusion is an enhancement of image-to-image diffusion models that adds the ability to control the amount of change applied to each image fragment via a change map.

Dynamic R-CNN is an object detection method that adjusts the label assignment criteria (IoU threshold) and the shape of regression loss function (parameters of Smooth L1 Loss) automatically based on the statistics of proposals during training. The motivation is that in previous two-stage object detectors, there is an inconsistency problem between the fixed network settings and the dynamic training procedure. For example, the fixed label assignment strategy and regression loss function cannot fit the distribution change of proposals and thus are harmful to training high quality detectors. It consists of two components: Dynamic Label Assignment and Dynamic Smooth L1 Loss, which are designed for the classification and regression branches, respectively. For Dynamic Label Assignment, we want our model to be discriminative for high IoU proposals, so we gradually adjust the IoU threshold for positive/negative samples based on the proposals distribution in the training procedure. Specifically, we set the threshold as the IoU of the proposal at a certain percentage since it can reflect the quality of the overall distribution. For Dynamic Smooth L1 Loss, we want to change the shape of the regression loss function to adaptively fit the distribution change of error and ensure the contribution of high quality samples to training. This is achieved by adjusting the $\beta$ in Smooth L1 Loss based on the error distribution of the regression loss function, in which $\beta$ actually controls the magnitude of the gradient of small errors.

An EESP Unit, or Extremely Efficient Spatial Pyramid of Depth-wise Dilated Separable Convolutions, is an image model block designed for edge devices. It was proposed as part of the ESPNetv2 CNN architecture. This building block is based on a reduce-split-transform-merge strategy. The EESP unit first projects the high-dimensional input feature maps into low-dimensional space using groupwise pointwise convolutions and then learns the representations in parallel using depthwise dilated separable convolutions with different dilation rates. Different dilation rates in each branch allow the EESP unit to learn the representations from a large effective receptive field. To remove the gridding artifacts caused by dilated convolutions, the EESP fuses the feature maps using hierarchical feature fusion (HFF).

The Enhanced Fusion Framework proposes three different ideas to improve the existing MI-based BCI frameworks. Image source: Fumanal-Idocin et al.

Flow Alignment Module, or FAM, is a flow-based align module for scene parsing to learn Semantic Flow between feature maps of adjacent levels and broadcast high-level features to high resolution features effectively and efficiently. The concept of Semantic Flow is inspired from optical flow, which is widely used in video processing task to represent the pattern of apparent motion of objects, surfaces, and edges in a visual scene caused by relative motion. The authors postulate that the relatinship between two feature maps of arbitrary resolutions from the same image can also be represented with the “motion” of every pixel from one feature map to the other one. Once precise Semantic Flow is obtained, the network is able to propagate semantic features with minimal information loss. In the FAM module, the transformed high-resolution feature map are combined with the low-resolution feature map to generate the semantic flow field, which is utilized to warp the low-resolution feature map to high-resolution feature map.

FreeAnchor is an anchor supervision method for object detection. Many CNN-based object detectors assign anchors for ground-truth objects under the restriction of object-anchor Intersection-over-Unit (IoU). In contrast, FreeAnchor is a learning-to-match approach that breaks the IoU restriction, allowing objects to match anchors in a flexible manner. It updates hand-crafted anchor assignment to free anchor matching by formulating detector training as a maximum likelihood estimation (MLE) procedure. FreeAnchor targets at learning features which best explain a class of objects in terms of both classification and localization.

Graph Path Feature Learning is a probabilistic rule learner optimized to mine instantiated first-order logic rules from knowledge graphs. Instantiated rules contain constants extracted from KGs. Compared to abstract rules that contain no constants, instantiated rules are capable of explaining and expressing concepts in more detail. GPFL utilizes a novel two-stage rule generation mechanism that first generalizes extracted paths into templates that are acyclic abstract rules until a certain degree of template saturation is achieved, then specializes the generated templates into instantiated rules.

Gated Positional Self-Attention (GPSA) is a self-attention module for vision transformers, used in the ConViT architecture, that can be initialized as a convolutional layer -- helping a ViT learn inductive biases about locality.

Gated Transformer-XL, or GTrXL, is a Transformer-based architecture for reinforcement learning. It introduces architectural modifications that improve the stability and learning speed of the original Transformer and XL variant. Changes include: - Placing the layer normalization on only the input stream of the submodules. A key benefit to this reordering is that it now enables an identity map from the input of the transformer at the first layer to the output of the transformer after the last layer. This is in contrast to the canonical transformer, where there are a series of layer normalization operations that non-linearly transform the state encoding. - Replacing residual connections with gating layers. The authors' experiments found that GRUs were the most effective form of gating.

A Gated Convolutional Network is a type of language model that combines convolutional networks with a gating mechanism. Zero padding is used to ensure future context can not be seen. Gated convolutional layers can be stacked on top of other hierarchically. Model predictions are then obtained with an adaptive softmax layer.

GradDrop, or Gradient Sign Dropout, is a probabilistic masking procedure which samples gradients at an activation layer based on their level of consistency. It is applied as a layer in any standard network forward pass, usually on the final layer before the prediction head to save on compute overhead and maximize benefits during backpropagation. Below, we develop the GradDrop formalism. Throughout, o denotes elementwise multiplication after any necessary tiling operations (if any) are completed. To implement GradDrop, we first define the Gradient Positive Sign Purity, $\mathcal{P}$, as $$ \mathcal{P}=\frac{1}{2}\left(1+\frac{\sum\_{i} \nabla L_\{i}}{\sum\_{i}\left|\nabla L\_{i}\right|}\right) $$ $\mathcal{P}$ is bounded by $[0,1] .$ For multiple gradient values $\nabla\_{a} L\_{i}$ at some scalar $a$, we see that $\mathcal{P}=0$ if $\nabla_{a} L\_{i}<0 $ $\forall i$, while $\mathcal{P}=1$ if $\nabla\_{a} L\_{i}>0$ $\forall i $. Thus, $\mathcal{P}$ is a measure of how many positive gradients are present at any given value. We then form a mask for each gradient $\mathcal{M}\_{i}$ as follows: $$ \mathcal{M}\_{i}=\mathcal{I}[f(\mathcal{P})>U] \circ \mathcal{I}\left[\nabla L\_{i}>0\right]+\mathcal{I}[f(\mathcal{P})<U] \circ \mathcal{I}\left[\nabla L\_{i}<0\right] $$ for $\mathcal{I}$ the standard indicator function and $f$ some monotonically increasing function (often just the identity) that maps $[0,1] \mapsto[0,1]$ and is odd around $(0.5,0.5)$. $U$ is a tensor composed of i.i.d $U(0,1)$ random variables. The $\mathcal{M}\_{i}$ is then used to produce a final gradient $\sum \mathcal{M}\_{i} \nabla L\_{i}$

GridMask is a data augmentation method that randomly removes some pixels of an input image. Unlike other methods, the region that the algorithm removes is neither a continuous region nor random pixels in dropout. Instead, the algorithm removes a region with disconnected pixel sets, as shown in the Figure. We express the setting as $$ \tilde{\mathbf{x}}=\mathbf{x} \times M $$ where $\mathbf{x} \in R^{H \times W \times C}$ represents the input image, $M \in$ $\{0,1\}^{H \times W}$ is the binary mask that stores pixels to be removed, and $\tilde{\mathbf{x}} \in R^{H \times W \times C}$ is the result produced by the algorithm. For the binary mask $M$, if $M_{i, j}=1$ we keep pixel $(i, j)$ in the input image; otherwise we remove it. GridMask is applied after the image normalization operation. The shape of $M$ looks like a grid, as shown in the Figure . Four numbers $\left(r, d, \delta_{x}, \delta_{y}\right)$ are used to represent a unique $M$. Every mask is formed by tiling the units. $r$ is the ratio of the shorter gray edge in a unit. $d$ is the length of one unit. $\delta\_{x}$ and $\delta\_{y}$ are the distances between the first intact unit and boundary of the image.

Guided Anchoring is an anchoring scheme for object detection which leverages semantic features to guide the anchoring. The method is motivated by the observation that objects are not distributed evenly over the image. The scale of an object is also closely related to the imagery content, its location and geometry of the scene. Following this intuition, the method generates sparse anchors in two steps: first identifying sub-regions that may contain objects and then determining the shapes at different locations.

A Hermite Activations is a type of activation function which uses a smooth finite Hermite polynomial base as a substitute for non-smooth ReLUs. Relevant Paper: Lokhande et al

A Hopfield Layer is a module that enables a network to associate two sets of vectors. This general functionality allows for transformer-like self-attention, for decoder-encoder attention, for time series prediction (maybe with positional encoding), for sequence analysis, for multiple instance learning, for learning with point sets, for combining data sources by associations, for constructing a memory, for averaging and pooling operations, and for many more. In particular, the Hopfield layer can readily be used as plug-in replacement for existing layers like pooling layers (max-pooling or average pooling, permutation equivariant layers, GRU & LSTM layers, and attention layers. The Hopfield layer is based on modern Hopfield networks with continuous states that have very high storage capacity and converge after one update.

I-BERT is a quantized version of BERT that quantizes the entire inference with integer-only arithmetic. Based on lightweight integer only approximation methods for nonlinear operations, e.g., GELU, Softmax, and Layer Normalization, it performs an end-to-end integer-only BERT inference without any floating point calculation. In particular, GELU and Softmax are approximated with lightweight second-order polynomials, which can be evaluated with integer-only arithmetic. For LayerNorm, integer-only computation is performed by leveraging a known algorithm for integer calculation of square root.

Implicit Subspace Prior Learning, or ISPL, is a framework to approach dual-blind face restoration, with two major distinctions from previous restoration methods: 1) Instead of assuming an explicit degradation function between LQ and HQ domain, it establishes an implicit correspondence between both domains via a mutual embedding space, thus avoid solving the pathological inverse problem directly. 2) A subspace prior decomposition and fusion mechanism to dynamically handle inputs at varying degradation levels with consistent high-quality restoration results.

IoU-Net is an object detection architecture that introduces localization confidence. IoU-Net learns to predict the IoU between each detected bounding box and the matched ground-truth. The network acquires this confidence of localization, which improves the NMS procedure by preserving accurately localized bounding boxes. Furthermore, an optimization-based bounding box refinement method is proposed, where the predicted IoU is formulated as the objective.

Kernel Inducing Points, or KIP, is a meta-learning algorithm for learning datasets that can mitigate the challenges which occur for naturally occurring datasets without a significant sacrifice in performance. KIP uses kernel-ridge regression to learn $\epsilon$-approximate datasets. It can be regarded as an adaption of the inducing point method for Gaussian processes to the case of Kernel Ridge Regression.

Graph structure is learnable

Leverage learning suggests that it is possible to strategically use minimal task-specific data to enhance task-specific capabilities, while non-specific capabilities can be learned from more general data.

MPRNet is a multi-stage progressive image restoration architecture that progressively learns restoration functions for the degraded inputs, thereby breaking down the overall recovery process into more manageable steps. Specifically, the model first learns the contextualized features using encoder-decoder architectures and later combines them with a high-resolution branch that retains local information. At each stage, a per-pixel adaptive design is introduced that leverages in-situ supervised attention to reweight the local features.

Meena is a multi-turn open-domain chatbot trained end-to-end on data mined and filtered from public domain social media conversations. This 2.6B parameter neural network is simply trained to minimize perplexity of the next token. A seq2seq model is used with the Evolved Transformer as the main architecture. The model is trained on multi-turn conversations where the input sequence is all turns of the context and the output sequence is the response.

MixText is a semi-supervised learning method for text classification, which uses a new data augmentation method called TMix. TMix creates a large amount of augmented training samples by interpolating text in hidden space. The technique leverages advances in data augmentation to guess low-entropy labels for unlabeled data, making them as easy to use as labeled data.

The NA method can be divided into two steps: (i) Training a neural network approximation of f , and (ii) inference of xˆ. Step (i) is conventional and involves training a generic neural network on a dataset ˆ of input/output pairs from the simulator, denoted D, resulting in f, an approximation of the forward ˆ model. This is illustrated in the left inset of Fig 1. In step (ii), our goal is to use ∂f/∂x to help us gradually adjust x so that we achieve a desired output of the forward model, y. This is similar to many classical inverse modeling approaches, such as the popular Adjoint method [8, 9]. For many practical ˆ expression for the simulator, from which it is trivial to compute ∂f/∂x, and furthermore, we can use modern deep learning software packages to efficiently estimate gradients, given a loss function L. More formally, let y be our target output, and let xˆi be our current estimate of the solution, where i indexes each solution we obtain in an iterative gradient-based estimation procedure. Then we compute xˆi+1 with inverse problems, however, obtaining ∂f/∂x requires significant expertise and/or effort, making these approaches challenging. Crucially, fˆ from step (i) provides us with a closed-form differentiable

OPT-IML is a version of OPT fine-tuned on a large collection of 1500+ NLP tasks divided into various task categories.

Online Normalization is a normalization technique for training deep neural networks. To define Online Normalization. we replace arithmetic averages over the full dataset in with exponentially decaying averages of online samples. The decay factors $\alpha\_{f}$ and $\alpha\_{b}$ for forward and backward passes respectively are hyperparameters for the technique. We allow incoming samples $x\_{t}$, such as images, to have multiple scalar components and denote feature-wide mean and variance by $\mu\left(x\_{t}\right)$ and $\sigma^{2}\left(x\_{t}\right)$. The algorithm also applies to outputs of fully connected layers with only one scalar output per feature. In fact, this case simplifies to $\mu\left(x\_{t}\right) = x\_{t}$ and $\sigma\left(x\_{t}\right) = 0$. Denote scalars $\mu\_{t}$ and $\sigma\_{t}$ to denote running estimates of mean and variance across all samples. The subscript $t$ denotes time steps corresponding to processing new incoming samples. Online Normalization uses an ongoing process during the forward pass to estimate activation means and variances. It implements the standard online computation of mean and variance generalized to processing multi-value samples and exponential averaging of sample statistics. The resulting estimates directly lead to an affine normalization transform. $$ y\_{t} = \frac{x\_{t} - \mu\_{t-1}}{\sigma\_{t-1}} $$ $$ \mu\_{t} = \alpha\_{f}\mu\_{t-1} + \left(1-\alpha\_{f}\right)\mu\left(x\_{t}\right) $$ $$ \sigma^{2}\_{t} = \alpha\_{f}\sigma^{2}\_{t-1} + \left(1-\alpha\_{f}\right)\sigma^{2}\left(x\_{t}\right) + \alpha\_{f}\left(1-\alpha\_{f}\right)\left(\mu\left(x\_{t}\right) - \mu\_{t-1}\right)^{2} $$

Probabilistically Masked Language Model, or PMLM, is a type of language model that utilizes a probabilistic masking scheme, aiming to bridge the gap between masked and autoregressive language models. The basic idea behind the connection of two categories of models is similar to MADE by Germain et al (2015). PMLM is a masked language model with a probabilistic masking scheme, which defines the way sequences are masked by following a probabilistic distribution. The authors employ a simple uniform distribution of the masking ratio and name the model as u-PMLM.

Primal Wasserstein Imitation Learning, or PWIL, is a method for imitation learning which ties to the primal form of the Wasserstein distance between the expert and the agent state-action distributions. The reward function is derived offline, as opposed to recent adversarial IL algorithms that learn a reward function through interactions with the environment, and requires little fine-tuning.