Contrastive Language-Image Pre-training (CLIP) is a technique for training a pair of neural network models, one for image understanding and one for text understanding, using a contrastive objective. This method has enabled broad applications across multiple domains, including cross-modal retrieval, text-to-image generation, and aesthetic ranking. == Algorithm == The CLIP method trains a pair of models contrastively. One model takes in a piece of text as input and outputs a single vector representing its semantic content. The other model takes in an image and similarly outputs a single vector representing its visual content. The models are trained so that the vectors corresponding to semantically similar text-image pairs are close together in the shared vector space, while those corresponding to dissimilar pairs are far apart. To train a pair of CLIP models, one would start by preparing a large dataset of image-caption pairs. During training, the models are presented with batches of N {\displaystyle N} image-caption pairs. Let the outputs from the text and image models be respectively v 1 , . . . , v N , w 1 , . . . , w N {\displaystyle v_{1},...,v_{N},w_{1},...,w_{N}} . Two vectors are considered "similar" if their dot product is large. The loss incurred on this batch is the multi-class N-pair loss, which is a symmetric cross-entropy loss over similarity scores: − 1 N ∑ i ln e v i ⋅ w i / T ∑ j e v i ⋅ w j / T − 1 N ∑ j ln e v j ⋅ w j / T ∑ i e v i ⋅ w j / T {\displaystyle -{\frac {1}{N}}\sum _{i}\ln {\frac {e^{v_{i}\cdot w_{i}/T}}{\sum _{j}e^{v_{i}\cdot w_{j}/T}}}-{\frac {1}{N}}\sum _{j}\ln {\frac {e^{v_{j}\cdot w_{j}/T}}{\sum _{i}e^{v_{i}\cdot w_{j}/T}}}} In essence, this loss function encourages the dot product between matching image and text vectors ( v i ⋅ w i {\displaystyle v_{i}\cdot w_{i}} ) to be high, while discouraging high dot products between non-matching pairs. The parameter T > 0 {\displaystyle T>0} is the temperature, which is parameterized in the original CLIP model as T = e − τ {\displaystyle T=e^{-\tau }} where τ ∈ R {\displaystyle \tau \in \mathbb {R} } is a learned parameter. Other loss functions are possible. For example, Sigmoid CLIP (SigLIP) proposes the following loss function: L = 1 N ∑ i , j ∈ 1 : N f ( ( 2 δ i , j − 1 ) ( e τ w i ⋅ v j + b ) ) {\displaystyle L={\frac {1}{N}}\sum _{i,j\in 1:N}f((2\delta _{i,j}-1)(e^{\tau }w_{i}\cdot v_{j}+b))} where f ( x ) = ln ( 1 + e − x ) {\displaystyle f(x)=\ln(1+e^{-x})} is the negative log sigmoid loss, and the Dirac delta symbol δ i , j {\displaystyle \delta _{i,j}} is 1 if i = j {\displaystyle i=j} else 0. == CLIP models == While the original model was developed by OpenAI, subsequent models have been trained by other organizations as well. === Image model === The image encoding models used in CLIP are typically vision transformers (ViT). The naming convention for these models often reflects the specific ViT architecture used. For instance, "ViT-L/14" means a "vision transformer large" (compared to other models in the same series) with a patch size of 14, meaning that the image is divided into 14-by-14 pixel patches before being processed by the transformer. The size indicator ranges from B, L, H, G (base, large, huge, giant), in that order. Other than ViT, the image model is typically a convolutional neural network, such as ResNet (in the original series by OpenAI), or ConvNeXt (in the OpenCLIP model series by LAION). Since the output vectors of the image model and the text model must have exactly the same length, both the image model and the text model have fixed-length vector outputs, which in the original report is called "embedding dimension". For example, in the original OpenAI model, the ResNet models have embedding dimensions ranging from 512 to 1024, and for the ViTs, from 512 to 768. Its implementation of ViT was the same as the original one, with one modification: after position embeddings are added to the initial patch embeddings, there is a LayerNorm. Its implementation of ResNet was the same as the original one, with 3 modifications: In the start of the CNN (the "stem"), they used three stacked 3x3 convolutions instead of a single 7x7 convolution, as suggested by. There is an average pooling of stride 2 at the start of each downsampling convolutional layer (they called it rect-2 blur pooling according to the terminology of ). This has the effect of blurring images before downsampling, for antialiasing. The final convolutional layer is followed by a multiheaded attention pooling. ALIGN a model with similar capabilities, trained by researchers from Google used EfficientNet, a kind of convolutional neural network. === Text model === The text encoding models used in CLIP are typically Transformers. In the original OpenAI report, they reported using a Transformer (63M-parameter, 12-layer, 512-wide, 8 attention heads) with lower-cased byte pair encoding (BPE) with 49152 vocabulary size. Context length was capped at 76 for efficiency. Like GPT, it was decoder-only, with only causally-masked self-attention. Its architecture is the same as GPT-2. Like BERT, the text sequence is bracketed by two special tokens [SOS] and [EOS] ("start of sequence" and "end of sequence"). Take the activations of the highest layer of the transformer on the [EOS], apply LayerNorm, then a final linear map. This is the text encoding of the input sequence. The final linear map has output dimension equal to the embedding dimension of whatever image encoder it is paired with. These models all had context length 77 and vocabulary size 49408. ALIGN used BERT of various sizes. == Dataset == === WebImageText === The CLIP models released by OpenAI were trained on a dataset called "WebImageText" (WIT) containing 400 million pairs of images and their corresponding captions scraped from the internet. The total number of words in this dataset is similar in scale to the WebText dataset used for training GPT-2, which contains about 40 gigabytes of text data. The dataset contains 500,000 text-queries, with up to 20,000 (image, text) pairs per query. The text-queries were generated by starting with all words occurring at least 100 times in English Wikipedia, then extended by bigrams with high mutual information, names of all Wikipedia articles above a certain search volume, and WordNet synsets. The dataset is private and has not been released to the public, and there is no further information on it. ==== Data preprocessing ==== For the CLIP image models, the input images are preprocessed by first dividing each of the R, G, B values of an image by the maximum possible value, so that these values fall between 0 and 1, then subtracting by [0.48145466, 0.4578275, 0.40821073], and dividing by [0.26862954, 0.26130258, 0.27577711]. The rationale was that these are the mean and standard deviations of the images in the WebImageText dataset, so this preprocessing step roughly whitens the image tensor. These numbers slightly differ from the standard preprocessing for ImageNet, which uses [0.485, 0.456, 0.406] and [0.229, 0.224, 0.225]. If the input image does not have the same resolution as the native resolution (224×224 for all except ViT-L/14@336px, which has 336×336 resolution), then the input image is first scaled by bicubic interpolation, so that its shorter side is the same as the native resolution, then the central square of the image is cropped out. === Others === ALIGN used over one billion image-text pairs, obtained by extracting images and their alt-tags from online crawling. The method was described as similar to how the Conceptual Captions dataset was constructed, but instead of complex filtering, they only applied a frequency-based filtering. Later models trained by other organizations had published datasets. For example, LAION trained OpenCLIP with published datasets LAION-400M, LAION-2B, and DataComp-1B. == Training == In the original OpenAI CLIP report, they reported training 5 ResNet and 3 ViT (ViT-B/32, ViT-B/16, ViT-L/14). Each was trained for 32 epochs. The largest ResNet model took 18 days to train on 592 V100 GPUs. The largest ViT model took 12 days on 256 V100 GPUs. All ViT models were trained on 224×224 image resolution. The ViT-L/14 was then boosted to 336×336 resolution by FixRes, resulting in a model. They found this was the best-performing model. In the OpenCLIP series, the ViT-L/14 model was trained on 384 A100 GPUs on the LAION-2B dataset, for 160 epochs for a total of 32B samples seen. == Applications == === Cross-modal retrieval === CLIP's cross-modal retrieval enables the alignment of visual and textual data in a shared latent space, allowing users to retrieve images based on text descriptions and vice versa, without the need for explicit image annotations. In text-to-image retrieval, users input descriptive text, and CLIP retrieves images with matching embeddings. In image-to-text retrieval, images are used to find related text content. CLIP’s ability to connect vis
Roposo
Roposo is an Indian video-sharing social media service, owned by Glance, a subsidiary of InMobi. Roposo provides a space where users can share posts related to different topics like food, comedy, music, poetry, fashion and travel. It is a platform where people express visually with homemade videos and photos. The app offers a TV-like browsing experience with user-generated content on its channels. Users can also use editing tools on the platform and upload their content. == History == Established in July 2014 under Relevant E-solutions Pvt. Ltd., Roposo is the brainchild of three IIT Delhi alumni – Mayank Bhangadia, Avinash Saxena, and Kaushal Shubhank. Under Bhangadia's leadership, the company pivoted from a fashion-based network into a short-form video platform with AI-powered moderation, and its journey was featured as a Harvard Business Publishing case study. In November 2019, Roposo was acquired by InMobi's Glance Digital Experience Pvt. Ltd.(the mobile content platform and part of the InMobi Group). When the Chinese-owned video-sharing app TikTok was banned on 30 June 2020, the app saw a huge spike in users with several TikTok users registering on Roposo. == Technology == The open platform has some features such as a TV-like browsing, different channels, a chat feature that lets buyers and sellers converse directly through the platform, and creation tools such as an option to add voice-over, music and GIF stickers for videos and photos.
ArcObjects
ArcObjects is a development environment of the ArcGIS family of applications. Using Visual Basic for Applications, C# or Java SDK for ArcGIS, it allows developers to extend these applications.ArcObjects is a library of Component Object Model (COM) components that build up the foundation of Esri's ArcGIS platform. ArcObjects is written primarily in the C++ programming language. Since ArcGIS is completely built on top of ArcObjects, the ArcGIS platform can be fully customized and extended by making use of its COM services and capabilities. This allows for easy extension of the ArcObjects data model with any programming language that is compatible with COM, such as Visual Basic, C#, Visual Basic.NET, Java and Python. COM enables components to be reused at a binary level, meaning developers do not require access to the source code of ArcObjects in order to extend the ArcGIS platform. For this reason, an ArcObjects programmer can make use of any type inside the ArcObjects system without knowing the implementation details of the type, only needing to know what the type is able to do. The ArcObjects data model is based on the COM standard, which makes it compatible with other COM objects and applications. This allows for easy integration and collaboration with other systems that are also based on the COM standard. The ArcGIS platform was built using ArcObjects types, such as classes, interfaces, and enumerations. ArcObjects use COM interfaces to organize and communicate properties and methods of its classes, ensuring compatibility with other COM-based objects and systems. When working with an ArcObjects COM class, its properties and methods are accessed solely through one of its implemented interfaces via the process of Query Interface (QI). Multiple interfaces are commonly available for classes in ArcObjects. For example, it is possible to query for additional interfaces implemented by an object after instantiation via the process of QI. Although only one interface can be used when instantiating an object, multiple interfaces are often available for classes in ArcObjects, allowing for greater flexibility and compatibility with other systems based on the COM standard.
Thinkfree Office
Thinkfree Office is a web-based commercial office productivity suite developed by South Korea-based Thinkfree Inc. It includes Word (a word processor), Spreadsheet (a spreadsheet) and Presentation (a presentation program). They are compatible with Microsoft Office's Word, PowerPoint, and Excel. It also features collaborative editing. The product is hosted on the client's server. == Supported file formats == Thinkfree Office supports ISO/IEC international standard ISO/IEC 26300 Open Document Format for Office Applications (odf, odt, odp, ods, odg). It also supports Microsoft's XML formats (docx, pptx, xlsx) and Microsoft's legacy binary formats (doc, ppt, xls). == Naming == The software was previously marketed under different names, such as Thinkfree Server, Thinkfree Online, Hancom Office Online, and Hancom Office Web. Eventually, the brand was consolidated under the name Thinkfree Office. == History == In June 2000, Thinkfree Inc. released Thinkfree Office, based in Silicon Valley, California. It is recognized as the world's first online office editor (predating Google Docs and Microsoft 365) and attracted significant media coverage, including reports on CNN. In 2001, Microsoft CEO Steve Ballmer highlighted Thinkfree as a significant competitor in a magazine interview, considering it a potential threat to his company, second only to Linux. In November 2003, Hancom, a South Korean office software company, signed a memorandum of understanding and subsequently acquired Thinkfree. In January 2004, Thinkfree expanded into other foreign markets. Subsidiary Haansoft USA, Inc. was created in San Jose, California to begin formal commercial operations in the US market. At the same time, a partnership was established with Riverdeep with the purpose of improving marketshare. In February 2004, expansion into the Japanese market began. A commercial agency agreement was signed with PSI in Shinjuku, Japan, which allowed for localized distribution. In addition, a global agreement was entered into with Yamada Denki, one of the three main computer distributors in Japan, for a total of 180,000 units. In May 2006, Thinkfree Office received the "Product of the Year" award at the Well-Connected Awards, USA. In January 2009, Thinkfree Mobile was launched at CES 2009 in Las Vegas. In April 2009, Thinkfree Live, Korea's first web office service, was launched. In June 2018, a partnership was formed with Amazon Web Services to integrate Thinkfree Office into WorkDocs, an in-house office suite. In October 2023, Hancom split its online office business unit as "Thinkfree Inc.".
Cloud management
Cloud management refers to the administration and oversight of cloud computing products and services. Public clouds are managed by cloud service providers, which operate the underlying infrastructure such as servers, storage, networking, and data center facilities. Users may also opt to manage their public cloud services with a third-party cloud management tool. Users of public cloud services can generally select from three basic cloud provisioning categories: User self-provisioning: Customers purchase cloud services directly from the provider, typically through a web form or console interface. The customer pays on a per-transaction basis. Advanced provisioning: Customers contract in advance a predetermined amount of resources, which are prepared in advance of service. The customer pays a flat fee or a monthly fee. Dynamic provisioning: The provider allocates resources when the customer needs them, then decommissions them when they are no longer needed. The customer is charged on a pay-per-use basis. Managing a private cloud requires software tools to help create a virtualized pool of compute resources, provide a self-service portal for end users and handle security, resource allocation, tracking and billing. Management tools for private clouds tend to be service driven, as opposed to resource driven, because cloud environments are typically highly virtualized and organized in terms of portable workloads. In hybrid cloud environments, compute, network and storage resources must be managed across multiple domains, so a good management strategy should start by defining what needs to be managed, and where and how to do it. Policies to help govern these domains should include configuration and installation of images, access control, and budgeting and reporting. Access control often includes the use of Single sign-on (SSO), in which a user logs in once and gains access to all systems without being prompted to log in again at each of them. == Characteristics of Cloud Management == Cloud management combines software and technologies in a design for managing cloud environments. Software developers have responded to the management challenges of cloud computing with a variety of cloud management platforms and tools. These tools include native tools offered by public cloud providers as well as third-party tools designed to provide consistent functionality across multiple cloud providers. Administrators must balance the competing requirements of efficient consistency across different cloud platforms with access to different native functionality within individual cloud platforms. The growing acceptance of public cloud and increased multicloud usage is driving the need for consistent cross-platform management. Rapid adoption of cloud services is introducing a new set of management challenges for those technical professionals responsible for managing IT systems and services. Cloud-management platforms and tools should have the ability to provide minimum functionality in the following categories. Functionality can be both natively provided or orchestrated via third-party integration. Provisioning and orchestration: create, modify, and delete resources as well as orchestrate workflows and management of workloads Automation: Enable cloud consumption and deployment of app services via infrastructure-as-code and other DevOps concepts Security and compliance: manage role-based access of cloud services and enforce security configurations Service request: collect and fulfill requests from users to access and deploy cloud resources. Monitoring and logging: collect performance and availability metrics as well as automate incident management and log aggregation Inventory and classification: discover and maintain pre-existing brownfield cloud resources plus monitor and manage changes Cost management and optimization: track and rightsize cloud spend and align capacity and performance to actual demand Migration, backup, and DR: enable data protection, disaster recovery, and data mobility via snapshots and/or data replication Organizations may group these criteria into key use cases including Cloud Brokerage, DevOps Automation, Governance, and Day-2 Life Cycle Operations. Enterprises with large-scale cloud implementations may require more robust cloud management tools which include specific characteristics, such as the ability to manage multiple platforms from a single point of reference, or intelligent analytics to automate processes like application lifecycle management. High-end cloud management tools should also have the ability to handle system failures automatically with capabilities such as self-monitoring, an explicit notification mechanism, and include failover and self-healing capabilities. == Multi-Cloud and Hybrid Cloud Management Challenges == Legacy management infrastructures, which are based on the concept of dedicated system relationships and architecture constructs, are not well suited to cloud environments where instances are continually launched and decommissioned. Instead, the dynamic nature of cloud computing requires monitoring and management tools that are adaptable, extensible and customizable. Cloud computing presents a number of management challenges. Companies using public clouds do not have ownership of the equipment hosting the cloud environment, and because the environment is not contained within their own networks, public cloud customers do not have full visibility or control. Users of public cloud services must also integrate with an architecture defined by the cloud provider, using its specific parameters for working with cloud components. Integration includes tying into the cloud APIs for configuring IP addresses, subnets, firewalls and data service functions for storage. Because control of these functions is based on the cloud provider’s infrastructure and services, public cloud users must integrate with the cloud infrastructure management. Capacity management is a challenge for both public and private cloud environments because end users have the ability to deploy applications using self-service portals. Applications of all sizes may appear in the environment, consume an unpredictable amount of resources, then disappear at any time. A possible solution is profiling the applications impact on computational resources. As result, the performance models allow the prediction of how resource utilization changes according to application patterns. Thus, resources can be dynamically scaled to meet the expected demand. This is critical to cloud providers that need to provision resources quickly to meet a growing demand by their applications. Charge-back—or, pricing resource use on a granular basis—is a challenge for both public and private cloud environments. Charge-back is a challenge for public cloud service providers because they must price their services competitively while still creating profit. Users of public cloud services may find charge-back challenging because it is difficult for IT groups to assess actual resource costs on a granular basis due to overlapping resources within an organization that may be paid for by an individual business unit, such as electrical power. For private cloud operators, charge-back is fairly straightforward, but the challenge lies in guessing how to allocate resources as closely as possible to actual resource usage to achieve the greatest operational efficiency. Exceeding budgets can be a risk. Hybrid cloud environments, which combine public and private cloud services, sometimes with traditional infrastructure elements, present their own set of management challenges. These include security concerns if sensitive data lands on public cloud servers, budget concerns around overuse of storage or bandwidth and proliferation of mismanaged images. Managing the information flow in a hybrid cloud environment is also a significant challenge. On-premises clouds must share information with applications hosted off-premises by public cloud providers, and this information may change constantly. Hybrid cloud environments also typically include a complex mix of policies, permissions and limits that must be managed consistently across both public and private clouds. == Cloud Management Platforms (CMP) == CMPs provide a means for a cloud service customer to manage the deployment and operation of applications and associated datasets across multiple cloud service infrastructures, including both on-premises cloud infrastructure and public cloud service provider infrastructure. In other words, CMPs provide management capabilities for hybrid cloud and multi-cloud environments. A cloud management platform (CMP) provides broad cloud management functionality atop both public cloud provider platforms and private cloud platforms. CMPs manage cloud services and resources that are distributed across multiple cloud platforms. The value of CMPs stands in delivering the maximum level of consistency between platforms without comp
Self-supervised learning
Self-supervised learning (SSL) is a paradigm in machine learning where a model is trained on a task using the data itself to generate supervisory signals, rather than relying on externally-provided labels. In the context of neural networks, self-supervised learning aims to leverage inherent structures or relationships within the input data to create meaningful training signals. SSL tasks are designed so that solving them requires capturing essential features or relationships in the data. The input data is typically augmented or transformed in a way that creates pairs of related samples, where one sample serves as the input, and the other is used to formulate the supervisory signal. This augmentation can involve introducing noise, cropping, rotation, or other transformations. Self-supervised learning more closely imitates the way humans learn to classify objects. During SSL, the model learns in two steps. First, the task is solved based on an auxiliary or pretext classification task using pseudo-labels, which help to initialize the model parameters. Next, the actual task is performed with supervised or unsupervised learning. Self-supervised learning has produced promising results in recent years, and has found practical application in fields such as audio processing, and is being used by Facebook and others for speech recognition. == Pseudo-labels == Pseudo-labels are automatically generated labels that a model assigns to unlabeled data based on its own predictions. They are widely used in self-supervised and semi-supervised learning, where ground-truth annotations are limited or unavailable. By treating predicted labels as surrogate ground truth, learning algorithms can make use of large quantities of unlabeled data in the training process. Pseudo-labeling also plays an important role in systems that must adapt to concept drift, where the statistical properties of the data change over time. In these scenarios, the model may detect that an incoming instance deviates from previously learned behavior. The system then generates a classification result for that instance, and this predicted class is used as a pseudo-label for updating or retraining model components that are becoming outdated. This approach enables continuous adaptation in dynamic environments without requiring manual annotation. In many adaptive learning pipelines, pseudo-labels are chosen when the classifier produces sufficiently confident predictions, reducing the risk of propagating errors. These pseudo-labeled instances are then incorporated into training to refresh or evolve the model's understanding of emerging data patterns, particularly when existing components show signs of “aging” due to drift or distributional shifts. This strategy reduces reliance on manual labeling while helping maintain long-term model performance. == Types == === Autoassociative self-supervised learning === Autoassociative self-supervised learning is a specific category of self-supervised learning where a neural network is trained to reproduce or reconstruct its own input data. In other words, the model is tasked with learning a representation of the data that captures its essential features or structure, allowing it to regenerate the original input. The term "autoassociative" comes from the fact that the model is essentially associating the input data with itself. This is often achieved using autoencoders, which are a type of neural network architecture used for representation learning. Autoencoders consist of an encoder network that maps the input data to a lower-dimensional representation (latent space), and a decoder network that reconstructs the input from this representation. The training process involves presenting the model with input data and requiring it to reconstruct the same data as closely as possible. The loss function used during training typically penalizes the difference between the original input and the reconstructed output (e.g. mean squared error). By minimizing this reconstruction error, the autoencoder learns a meaningful representation of the data in its latent space. === Contrastive self-supervised learning === For a binary classification task, training data can be divided into positive examples and negative examples. Positive examples are those that match the target. For example, if training a classifier to identify birds, the positive training data would include images that contain birds. Negative examples would be images that do not. Contrastive self-supervised learning uses both positive and negative examples. The loss function in contrastive learning is used to minimize the distance between positive sample pairs, while maximizing the distance between negative sample pairs. An early example uses a pair of 1-dimensional convolutional neural networks to process a pair of images and maximize their agreement. Contrastive Language-Image Pre-training (CLIP) allows joint pretraining of a text encoder and an image encoder, such that a matching image-text pair have image encoding vector and text encoding vector that span a small angle (having a large cosine similarity). InfoNCE (Noise-Contrastive Estimation) is a method to optimize two models jointly, based on Noise Contrastive Estimation (NCE). Given a set X = { x 1 , … x N } {\displaystyle X=\left\{x_{1},\ldots x_{N}\right\}} of N {\displaystyle N} random samples containing one positive sample from p ( x t + k ∣ c t ) {\displaystyle p\left(x_{t+k}\mid c_{t}\right)} and N − 1 {\displaystyle N-1} negative samples from the 'proposal' distribution p ( x t + k ) {\displaystyle p\left(x_{t+k}\right)} , it minimizes the following loss function: L N = − E X [ log f k ( x t + k , c t ) ∑ x j ∈ X f k ( x j , c t ) ] {\displaystyle {\mathcal {L}}_{\mathrm {N} }=-\mathbb {E} _{X}\left[\log {\frac {f_{k}\left(x_{t+k},c_{t}\right)}{\sum _{x_{j}\in X}f_{k}\left(x_{j},c_{t}\right)}}\right]} === Non-contrastive self-supervised learning === Non-contrastive self-supervised learning (NCSSL) uses only positive examples. Counterintuitively, NCSSL converges on a useful local minimum rather than reaching a trivial solution, with zero loss. For the example of binary classification, it would trivially learn to classify each example as positive. Effective NCSSL requires an extra predictor on the online side that does not back-propagate on the target side. === Joint-Embedding and Predictive Architectures === A major class of self-supervised learning moves beyond contrastive pairs, instead maximizing the agreement between views while preventing collapse through statistical constraints. Rooted in Deep Canonical Correlation Analysis (Deep CCA), this approach includes Joint-Embedding Architectures (JEA) like Barlow Twins and VICReg, which enforce covariance constraints to learn invariant representations without negative sampling. Deep Latent Variable Path Modelling (DLVPM) generalizes this to multimodal systems, using path models to enforce correlation and orthogonality across diverse data types. In 2022 Yann LeCun introduced Joint-Embedding Predictive Architectures (JEPA) as a step towards decision making, reasoning, and autonomous human intelligence in machines, including self-improvement through autonomous learning. Founded in representation learning, LeCun included the concept of a “world model” in JEPA which aims to enable machines to replicate human intellect by providing machines with a concept for the world in which they exist. Unlike autoencoders, JEPAs operate entirely in latent space, avoiding pixel-level noise to focus on semantic structure. Rather than just learning invariance, JEPAs learn by predicting masked latent representations from visible context. JEPA has been applied to domains such as image analysis, audio processing, and motion in images and video. == Comparison with other forms of machine learning == SSL belongs to supervised learning methods insofar as the goal is to generate a classified output from the input. At the same time, however, it does not require the explicit use of labeled input-output pairs. Instead, correlations, metadata embedded in the data, or domain knowledge present in the input are implicitly and autonomously extracted from the data. These supervisory signals, extracted from the data, can then be used for training. SSL is similar to unsupervised learning in that it does not require labels in the sample data. Unlike unsupervised learning, however, learning is not done using inherent data structures. Semi-supervised learning combines supervised and unsupervised learning, requiring only a small portion of the learning data be labeled. In transfer learning, a model designed for one task is reused on a different task. Training an autoencoder intrinsically constitutes a self-supervised process, because the output pattern needs to become an optimal reconstruction of the input pattern itself. However, in current jargon, the term 'self-supervised' often refers to tasks based on a pretext-task training setup
Windows Live OneCare Safety Scanner
Windows Live OneCare Safety Scanner (formerly Windows Live Safety Center and codenamed Vegas) was an online scanning, PC cleanup, and diagnosis service to help remove of viruses, spyware/adware, and other malware. It was a free web service that was part of Windows Live. On November 18, 2008, Microsoft announced the discontinuation of Windows Live OneCare, offering users a new free anti-malware suite Microsoft Security Essentials, which had been available since the second half of 2009. However, Windows Live OneCare Safety Scanner, under the same branding as Windows Live OneCare, was not discontinued during that time. The service was officially discontinued on April 15, 2011 and replaced with Microsoft Safety Scanner. == Overview == Windows Live OneCare Safety Scanner offered a free online scanning and protection from threats. The Windows Live OneCare Safety Scanner must be downloaded and installed to your computer to scan your computer. The "Full Service Scan" looks for common PC health issues such as viruses, temporary files, and open network ports. It searches and removes viruses, improves a computer's performance, and removes unnecessary clutter on the PC's hard disk. The user can choose between a "Full Scan" (which can be customized) or a "Quick Scan". The "Full Scan" scans for viruses (comprehensive scan or quick scan), hard disk performance (Disk fragmentation scan and/or Desk cleanup scan) and network safety (open port scan). The "Quick Scan" only scans for viruses, only on specific areas on the computer. The quick scan is faster than the full scan, hence that appellation. The service also provides a virus database, information about online threats, and general computer security documentation and tools. == Limits == The virus scanner on the Windows Live OneCare Safety Scanner site runs a scan of the user's computer only when the site is visited. It does not run periodic scans of the system, and does not provide features to prevent viruses from infecting the computer at the time, or thereafter. It simply resolves detected infections. Many users who have posted on the Product Feedback forum report script errors relating to Internet Explorer 7 (besides IE being the only browser supported by this service). The OneCare safety scanner team have been actively solving these problems, many of them registry-related.