current location:Home > Technical Articles > Technology peripherals > AI
- Direction:
- All web3.0 Backend Development Web Front-end Database Operation and Maintenance Development Tools PHP Framework Daily Programming WeChat Applet Common Problem Other Tech CMS Tutorial Java System Tutorial Computer Tutorials Hardware Tutorial Mobile Tutorial Software Tutorial Mobile Game Tutorial
- Classify:
-
- Tsinghua's latest! RoadBEV: How to achieve road surface reconstruction under BEV?
- Original title: RoadBEV: RoadSurfaceReconstructioninBird'sEyeView Paper link: https://arxiv.org/pdf/2404.06605.pdf Code link: https://github.com/ztsrxh/RoadBEV Author affiliation: Tsinghua University, University of California, Berkeley Paper idea: Road surface conditions, especially geometric contours, greatly affect the driving capabilities of autonomous vehicles. Vision-based online road reconstruction is expected to capture road information in advance. Existing solutions such as monocular depth estimation and stereo vision estimation have their limitations. Nearest Bird's Eye View (BEV
- AI 616 2024-04-16 19:55:24
-
- Extremely long sequences, extremely fast: LASP sequence parallelism for a new generation of efficient large language models
- The AIxiv column is a column where this site publishes academic and technical content. In the past few years, the AIxiv column of this site has received more than 2,000 reports, covering top laboratories from major universities and companies around the world, effectively promoting academic exchanges and dissemination. If you have excellent work that you want to share, please feel free to contribute or contact us for reporting. Submission email: liyazhou@jiqizhixin.com; zhaoyunfeng@jiqizhixin.com. From the top international GPT-4128K and Claude200K to the domestic KimiChat that supports more than 2 million words of text, large language models (LLM) have invariably become popular in long context technology. When the world's most
- AI 1175 2024-04-16 18:49:01
-
- Sora comes to Adobe video editing software! The new version of Premiere Pro opens the era of AI editing
- Sora is going to be integrated into Adobe video editing software. In the newly released PremierPro concept demonstration, Adobe showed the results of its cooperation with OpenAI: in addition to the main lens, a B-roll auxiliary lens was completely generated by Sora. In addition to Sora, other popular AI video tools Runway and Pika will also be available as options. Runway is used similarly to Sora in the demonstration and can generate a new auxiliary shot. Pika can naturally extend existing lenses for several seconds. It should be emphasized that these features are still in the early preview and research stage, and it has not yet been revealed when they will be released. In comparison, Adobe’s own AI product Firefly
- AI 444 2024-04-16 15:20:01
-
- How IoT sensors and artificial intelligence are revolutionizing smart buildings
- Over the past few years, especially in the wake of the COVID-19 pandemic, construction management has faced continued growth and evolution as expectations on facilities managers change and sustainability needs expand. The shift from offices to more hybrid and flexible work environments has changed the way commercial buildings are used, requiring real-time visibility into building usage, occupancy trends and more. The ever-changing construction management landscape demonstrates solutions that quickly adapt to new and flexible environments while improving overall productivity and performance. Smart Buildings Evaluate your own facilities and improvement opportunities Smart buildings are a growing trend that have the potential to not only streamline operations, but also reduce costs and increase visibility for all. Leveraging technologies such as the Internet of Things (IoT), artificial intelligence (AI), and automation
- AI 866 2024-04-16 15:16:37
-
- Inspired by the first-order optimization algorithm, Peking University Lin Zhouchen's team proposed a design method for neural network architecture with universal approximation properties
- As the basis of deep learning technology, neural networks have achieved effective results in many application fields. In practice, network architecture can significantly affect learning efficiency. A good neural network architecture can incorporate prior knowledge of the problem, establish network training, and improve computing efficiency. Currently, classic network architecture design methods include manual design, neural network architecture search (NAS) [1], and optimization-based network design methods [2]. Artificially designed network architectures such as ResNet, etc.; neural network architecture search searches for the best network structure in the search space through search or reinforcement learning; a mainstream paradigm in optimization-based design methods is algorithm expansion (algorithmunrolling), which Methods usually have an explicit purpose
- AI 744 2024-04-15 18:22:15
-
- Js-pytorch: Opening up a new world of front-end + AI
- Hello everyone, my name is Xu Xiaoxi. Recently I discovered a very interesting framework on Github - js-pytorch. It allows the front-end to easily use JavaScript to run deep learning frameworks. As a senior front-end technology player, I would like to share this framework with you today. In the current field of artificial intelligence, deep learning technology has become a technology that has attracted much attention. PyTorch is one of the deep learning frameworks that has attracted much attention. This framework has an important development rate in the field of deep learning. js-pytorch leverages JavaScript technology to introduce the power of PyTorch, enabling it to be used and used in the JavaScript world
- AI 557 2024-04-15 16:40:55
-
- Let's talk about end-to-end and next-generation autonomous driving systems, as well as some misunderstandings about end-to-end autonomous driving?
- In the past month, due to some well-known reasons, I have had very intensive exchanges with various teachers and classmates in the industry. An inevitable topic in the exchange is naturally end-to-end and the popular Tesla FSDV12. I would like to take this opportunity to sort out some of my thoughts and opinions at this moment for your reference and discussion. How to define an end-to-end autonomous driving system, and what problems should be expected to be solved end-to-end? According to the most traditional definition, an end-to-end system refers to a system that inputs raw information from sensors and directly outputs variables of concern to the task. For example, in image recognition, CNN can be called end-to-end compared to the traditional feature extractor + classifier method. In autonomous driving tasks, input data from various sensors (camera/LiDAR
- AI 1033 2024-04-15 16:13:01
-
- Concerned that U.S. tech giants are manipulating the global AI market, British regulators launch multiple investigations
- The UK Competition and Markets Authority (CMA) recently expressed concern that US technology companies may manipulate the global AI market. The CMA is investigating some technology companies that dominate the development and operation of basic AI models (such as FM), and the methods of cooperation between them (such as OpenAI and Microsoft). The agency pointed out that these technology companies have established an "Internet network" consisting of more than 90 partners (shown in Figure 1), and have received investments from Google, Apple, Microsoft, Meta, Amazon, Nvidia, etc. The CMA is concerned that these investments could be used to create barriers to other companies entering the AI market. image001.jpg Figure 1: The CMA survey of the “Internet” formed by American technology giants was completed last year.
- AI 342 2024-04-15 16:07:21
-
- NeurlPS, the world's top AI conference, has begun accepting papers from high school students
- What’s the next step after taking the college entrance examination, writing a paper? Cultivating AI talents should start from childhood. This sentence seems less and less like a joke. On Friday, the news that the top academic conference NeurIPS opened a paper track for high school students detonated the artificial intelligence community. The news is quoted from a new announcement from the organizers of the conference. Please note that this is not a workshop, but a main conference: we sincerely invite high school students to submit research papers related to the social impact of machine learning, and their projects will be presented at NeurIPS+2024. The organizing committee will select some of the finalists to present their projects virtually and highlight their work on the NeurIPS homepage. In addition, the lead authors of up to five winning projects will be invited to attend Ne
- AI 1098 2024-04-15 16:04:01
-
- Musk's new work! Shocking release of Grok-1.5V multi-modal model: perfect integration of digital and physical worlds
- Musk’s first-generation multi-modal model Grok-1.5V is finally here! Grok1.5V claims to be able to connect the digital world and the physical world. In addition to text capabilities, it can handle a variety of visual information such as documents, icons, screenshots, and photos. Grok 1.5V will be available to all early testers and all Grok users soon. Jim Fan, a senior scientist at NVIDIA, discovered the "biggest highlight" of Grok-1.5V: In his view, the most interesting thing about Grok-1.5V is its potential to solve edge cases of autonomous driving. By using "thinking chains" that can help cars solve complex scenarios, reason with rules and counterfactuals, and explain their decisions, Grok-1.5V can bring perception to the next level.
- AI 888 2024-04-15 16:01:01
-
- The once popular InstantID has a new way to play: stylized image generation, which has been open source
- The original team of InstantID launched InstantStyle, a new method of style migration. Stylized image generation is often referred to as style transfer, and its goal is to generate images that are consistent in style with a reference image and consistent with the content of the original image. This technology is based on diffusion methods (such as LoRA) for batch training of the same style data and cannot be migrated to new styles. Or based on inversion operations (such as StyleAlign), by restoring the style image to latentnoise, the K and V obtained by forward propagation are used to replace the style image in generation. This method often degrades the generation style due to the inversion operation. Recently, InstantID
- AI 570 2024-04-15 15:58:20
-
- ReFT (Representation Fine-tuning): a new large language model fine-tuning technology that is better than PeFT
- ReFT (Representation Finetuning) is a breakthrough method that promises to redefine the way we fine-tune large language models. According to a recent (April) paper published on arxiv by researchers at Stanford University, ReFT is significantly different from traditional weight-based fine-tuning methods, providing a more efficient and effective way to adapt to these large-scale models to Adapt to new tasks and areas! Before introducing this paper, let's take a look at PeFT. Parameter Efficient Fine-Tuning PeFTParameterEfficientFine-Tuning (PEFT) is an efficient fine-tuning method for fine-tuning a small number or additional model parameters. Fine-tuning with traditional predictive networks
- AI 1193 2024-04-15 15:30:02
-
- Who says elephants can't dance! Reprogramming large language models to achieve timing prediction of cross-modal interactions | ICLR 2024
- Recently, researchers from Australia's Monash University, Ant Group, IBM Research and other institutions have explored the application of model reprogramming (model reprogramming) on large language models (LLMs), and proposed a new perspective: efficient reprogramming of large language models (LLMs). Language model for general time series forecasting system, namely Time-LLM framework. This framework can achieve high-precision and efficient predictions without modifying the language model, and can surpass traditional time series models in multiple data sets and prediction tasks, allowing LLMs to demonstrate outstanding performance when processing cross-modal time series data. Like an elephant dancing. Recently, the development of large language models in the field of general intelligence, the new method of "large model + time series/time data"
- AI 526 2024-04-15 15:20:02
-
- To cultivate and develop new productive forces in the financial field, the 'AI + Finance' forum was successfully held in Pudong New Area
- On April 12, under the guidance of the Shanghai Pudong New Area Science and Technology and Economic Committee, hosted by the Shanghai Pudong New Area Industrial Development Promotion Center, Shanghai Pudong New Area Investment Promotion Center II, Shanghai Lujiazui (Group) Co., Ltd., Shanghai Pudong Science and Technology Entrepreneurship Center, The "AI + Finance" forum hosted by China Construction Bank Shanghai Pudong Branch and our website (Shanghai) Technology Co., Ltd. was successfully held in Zhangjiang Science City and Digital Intelligence World. The forum aims to discuss the development trends and innovative applications of artificial intelligence and the financial industry, attracting nearly 400 guests from industry, academia and finance. At the meeting, participating experts conducted in-depth discussions on hot topics such as the application and development trends of AI+ finance. At the same time, the forum also showcased intelligent customer service, intelligent investment advisory, and facial recognition technology.
- AI 860 2024-04-15 13:43:10