CMPUT 466/566 Machine learning

「文智助手」全场景办公AI助理AI+OPC全流程实践方案(大赛作品) 阅读详情

[CMPUT 466/566, Fall 2024] Machine learning

Course Project Description

Objectives:

1.   [10 marks] The basic goal of the mini-project is for the student to gain first-hand experience in

formulating a task as a machine learning problem and have a rigorous practice of applying machine learning algorithms.

2.   [5 marks] The second goal (optional to undergrads) is to accomplish a non-trivial machine learning

project, such as replicating a recent top-tier machine learning publication (published at ICML, NeurIPS, ICLR, etc.), proposing new models, and empirically analyzing machine learning models in a significant  way. Replicating a paper published at an unknown or non-machine learning venue may not constitute a non-trivial project. The 5 marks count as bonus for undergraduates but are included within 100 total marks for graduate students.

Example non-trivial project: Debiasing with Sufficient Projection: A General Theoretical Framework for Vector Representations

Note that only one project is expected. A non-trivial project must also satisfy the basic requirements.

Team work

Collaboration for the course project is possible only if

1) all team members have already had first-hand experience,

2) they intend to do a non-trivial project, and

3) the team must have no more than three members.

The team has to apply in NOI before the NOI deadline. The application may be declined if any of the team members does not have adequate machine learning background.

If teamwork is approved, the team members (name, ID, and email) and individual contributions must be stated clearly in all submissions. All team members must upload the submissions to their own eClass assignments.

In case the submitted team project only satisfies the basic requirements without any non-trivial components, all team members will share a total of 10 marks.

Timeline and submissions

All due time in this section is in Edmonton time. Every submission has a free extension.

A project intended to satisfy the basic requirements only (10 marks) does not need to submit the notice of intent or a proposal. They only need to submit the final project, and the deadline is 12:30PM, Dec 10 (extended to 12:30PM, Dec 17).

A non-trivial project requires significantly more time than a project satisfying basic requirements only, so a significant amount of time has to be set for the project. It must follow the mandatory timeline:

○   Sep 19 (extended to Sep 24): Notice of intent

○    Oct 17 (extended to Oct 22): Proposal

○    Dec 10 (extended to Dec 17): Final report

All deadlines are due by 12:30PM (Edmonton time).

A student must decide early if to attempt a non-trivial project. If so, the student must send a notice of intent (NOI) on eClass by the deadline, which can be a message, a title, or a short description. The NOI will not be reviewed but is mandatory for a non-trivial project.

If several students intend to form. a group, the NOI must also include every team member’s name, email (ccid), and prior experience in machine learning (such as a short bio). The approval of teamwork will be  based on the students’ background and the intended topic.

●   Any team member may leave the team unilaterally and submit a basic project. The rest of the team may still attempt a non-trivial project.

●    If a team is dissolved but there is dispute on who still owns the non-triviality project, then all members of the team fall into the category of basic projects.

○    Example: A team of two students do not wish to work together, but each claims the

ownership of the non-trivial project. Then, both students are ineligible for the non-trivial part.

For the non-trivial project, the student is supposed to read literature and prepare experimental environments after NOI. By the proposal deadline, the student must submit a pdf proposal to eClass. The instructor will read the proposal and make a comment, especially on how non-trivial the proposal is.

Notice that an intended non-trivial project may not get all 15 marks or, if not satisfying the basic requirements, may not even get 10 marks.

Basic Requirements [10 marks]:

●    Formulating a task into a machine learning problem. The student CANNOT re-use any task in coding assignments (namely, house price and MNIST datasets) as the course project.

●    Implementing a training-validation-test infrastructure, with a systematic way of hyperparameter tuning. The meaning of “training,” “validation,” “test,” and “hyperparameter” will be clear very  soon.

●    Comparing at least three machine learning algorithms. In addition, include trivial baselines (if  possible). For example, a majority guess for k-category classification yields 1/k accuracy. The machine learning algorithms must be reasonable for solving the task, and differ in some way   (e.g., having different hyperparameters does not count as different machine learning algorithms).

General machine learning packages may be used for the course project. However, the student cannot use the codebase specific to the task at hand and run a few scripts like “sh run.sh” .

Requirements for a non-trivial project [5 marks]:

A non-trivial project could be either replicating a recent, sophisticated machine learning paper, proposing new models, or conducting empirical analysis of machine learning models in a significant way.

Typically, a non-trivial project involves a significant amount of literature reading, programming, and experimentation. A student would not expect any additional marks by trying some CNN/RNN models, or applying existing code base to a new task in a straightforward way. If a student seeks non-triviality marks by replicating a recent paper, the student should assume the code base of that paper does not exist.

If a student has doubts about how non-trivial the project is, the student may check how much mathematical and algorithmic formulation there is.

Final report submission:

The submission must contain a PDF report and the code to reproduce the results. (Non-complying file format will result in mark deduction.)

The code should be submitted by a zip file through eClass or through a Google Drive link. If using Google Drive link, the student should

●    Make the folder “ Readable” by any university member, but keep the link in their custody except for the submission

●   Sharing the folder with the instructor does not work, because the project may not be graded by the instructor

●    In the event that the link is not kept in the student’s custody (e.g., sharing the link with friends or publishing the link), the student knows that other people may plagiarize the project. If the project is indeed plagiarized by others, the student is liable for Unauthorized Collaboration under SAIP   4c.

Knowingly advising, encouraging, aiding or assisting another person, directly or indirectly, to commit any violation under this policy. [Student Academic Integrity Policy Appendix A, 4c]

The format of the report is flexible, but generally, the report should contain

●   A short introduction, describing the background of the task

●    Problem formulation (what is input, what is output, where did you get the dataset, number of samples, etc.)

●   Approaches and baselines (what are the hyperparameters of each approach/baseline, how do you tune them)?

●    Evaluation metric (what is the measure of success, is it the real goal of the task, or an approximation? If it’s an approximation, why is it a reasonable approximation?)

●    Results. (What is the result of the approaches? How is it compared with baselines? How do you interpret the results?)

●   The report should NOT contain code snippets or program outputs.

Grading criteria:

Basic requirements [10 marks]:

●    If the submission is not a machine learning problem, then 0 marks.

●   Otherwise, the grading starts from 10 points. If one or more of the above requirements are not fulfilled, it will result in mark deduction for one or a few points.

●    Presentation enters the mark in a multiplicative way. The factor is 1 be default, if the report is reasonably well written. If the presented content is not readable, then the project will get 0 marks.

Non-triviality [5 marks]: Marking will consider literature review, proposed approach, and experimentation.

Statement of Expectations for AI Use:

AI tools, including but not restricted to generative models and online translation models, are not allowed.

Note: AI-flavored writing demonstrates poor presentation skills. For example, the text is oftentimes grandiose but empty. This will result in a devastatingly low mark (including a mark of 0) by merit, regardless of whether   there is proof of using AI tools.

Tips:

1.   The course project only counts 10--15% of the total marks, and obviously, this course focuses more on math derivations than coding. It is more important to formulate a machine learning system in a rigorous way and complete the project in time than do a super fancy project (which may require too much work and has a risk of not being finished in the course timeline).

2.   In fact, many students sought minimal efforts to obtain non-triviality marks in the past, which is not

possible. In general, not many students got the 5 marks and students should not worry about it. Even  for graduate students, getting a 5-mark deduction is not a problem, because the letter grade cutoff will be adjusted accordingly and may be different from undergraduates.

3.   Using external general-purpose machine learning packages is allowed but should be acknowledged   (e.g., use libsvm to solve the task by a few lines of function call). However, using a code base directly related to your task is not allowed (e.g., download a Git Hub repo and only write a few lines of script like “sh run.sh”).

4.   There is no constraint on the number of pages of the course report. However, the length should reflect the substance of the project, and in a normal case, a few pages suffice. An over-lengthed report will not yield a higher mark. On the contrary, it shows poor presentation skills (and may lead to mark deduction). The report must be written in text with results organized in an appropriate form. (such as tables and figures). Python notebook, code snippets, and program output are not considered as a textual report.

5.   We will grade the course project in a lenient way. However, we do not accept mark negotiation. The basic requirements are clearly stated above. The instructor will adjudicate the degree of non-triviality based on the same criteria applied to all students.

船载 OpenHarmony AIS 航海工作台:告警与防碰撞预警系统 船载OpenHarmony AIS航海工作台的告警与防碰撞系统,集成十余类告警(超速、偏航、会遇、SOS等),通过网格空间索引实现高效邻近判定,保障实时性。告警管理器统一调度规则、去重、通知与日志,支持悬浮球、语音、跑马灯、弹窗多渠道触达,确保航行安全。SOS紧急通道独立运行,避免干扰。系统设计强调性能优化与多场景适配,为航海安全提供核心支撑。 阅读详情

相关推荐

AI 应用开发:四维架构落地实战——以 HR Agent 为例拆解感知·推理·记忆·执行【意图路由、规划编排、长期记忆、工具执行、面试题】

本文基于四维架构(感知·推理·记忆·执行),以HR Agent为例,系统拆解大模型应用从“聊天”到“生产可用”的落地路径。通过Spring AI实现意图路由、规划编排、RAG检索与工具调用,结合真实踩坑复盘,揭示架构化设计对安全、成本与可维护性的关键价值,为AI Agent工程化提供可复用的实战框架。

王大师企业官方博客 1318

CMPUT 466/566 (Fall 2024) Syllabus Machine Learning EssentialsHaskell

Fall2024) SyllabusMachineLearning●●With whomEmailOpendoorLililmouThursday,3-4PMATH4-08Zijun Wuzijun4CCIS1-160Tue(2-2:30PM)Online:NicolasRebstocknrebstocThu(3-3:30PM)Online:harutoOnlineYu Wangyu35。

gave3gift的博客 827

VulnHub靶机渗透实战(详细思路讲解)

使用kali中的namp工具扫描开放端口,这时kali会用一个端口号向目标主机的65535个端口发送TCP连接建立请求,回应了ack=1,syn=q的端口就是开放端口号,这里可以看到21,22,80,3306号端口是开放的。可以看到这里得到了很多哈希,带*号的是禁用的或者是空密码,将三个不带星号的密码复制下来,创建一个shell.txt文档,将三个密码复制进去。这里看到我们是没有成功进入这个网页的,应该是目录的问题,我们将目录修改一下,但是没有显示出来,可能是编码问题,再回到我们找到的源码。

weixin_65891791的博客 2549

AI 问数真的能替代 BI 报表吗?

AI问数赋能自然语言查询,提升临时分析效率,但无法替代传统报表。报表适用于稳定输出与持续监控,AI则擅长探索性提问。企业用好AI问数,关键在于夯实数据模型、统一指标口径、健全权限体系与结果校验机制。合理组合报表、仪表板与AI,才能实现高效、准确的业务洞察。

葡萄城技术团队博客 338

量化策略分享 | MA超进化:LLT低延迟趋势线

移动平均线(MA)是我们技术分析中常用的一种趋势跟踪指标,但在使用的时候你是否也会有这样的烦恼:交易信号延迟太久,或者交易信号太频繁了!延迟性和平滑性问题似乎是不可兼得的“鱼和熊掌”。 针对这个问题,广发证券基于通信领域的滤波器原理,在《短线择时策略研究之三:低延迟趋势线与交易性择时》中创造性构建了:LLT低延迟趋势线(Low-lag Trendline)。 LLT低延迟趋势线是一种二阶滤波器,阶数相对适宜,相比MA、EMA等均线指标能够降低延迟,同时增强在低频部分的输出信号。 当然,..

量化交易研究 5314

光伏发电预测(LSTM、CNN_LSTM和XGBoost回归模型,Python代码)

一旦结果与预测差距较大,可能存在光伏系统的问题,需要找出原因,然后采取适当措施来修复光伏系统并做出更好的决策。例如,根据准确的预测,光伏系统运营商可以平衡电力消耗,并将多余的电力储备用于紧急情况。然而,光伏系统的发电量估算是一个挑战,因为光伏系统的发电量受到天气条件的极大影响。整个项目文件夹(第一个文件夹为xgboost模型相关的包文件,xg.py是xgboost模型英文注释版本,xg_chinese.py是xgboost模型中文注释版本)表示空气的温度,通常以摄氏度(°C)或华氏度(°F)为单位。

2301_79699110的博客 3034

【从零理解 Spring AI · 12】MCP 的两种角色:做客户端还是做服务端

上一篇讲了 MCP 的概念:标准协议、三大角色、M+N 替代 M×N。

nice_lcj520的博客 614

ZYNQ裸机LWIP双网口实现

文章目录前言一、LWIP双网口二、使用步骤1.部分说明2.部分代码总结 前言 一、LWIP双网口 由于项目需要多通讯口,查阅资料发现lwip支持多网口操作。于是,就开搞了。 二、使用步骤 1.部分说明 利用xemac_add函数在网卡链表中添加一个网卡,即使用netif块重复xemac_add操作。 先用轮询的方式实现双网口,后期修改为中断模式 2.部分代码 导入官方例程LWip echo server, 修改main函数,增加一个netif块 #if LWIP_IPV6==0 #if L..

weixin_35865280的博客 1万+

ChatGPT、Perplexity、Gemini 与 AI Overviews 引用对比:谁的引用更可信

AI 引擎的引用机制差异很大:ChatGPT 以行内引用加 Sources 列表呈现来源,Perplexity 几乎逐句可溯源,Gemini 在回答中内嵌相关来源链接,AI Overviews 则把多源聚合答案与来源列表一起放在搜索结果顶部。本文从引用透明度、来源偏好、可干预性 3 个维度对比这 4 个引擎,并给出 GEO 的 3 点启示,帮企业在多引擎环境下提升被引用的概率。

淡然一笑 的专栏 273

雷赛控制卡全方位使用指南:从硬件连接到高级应用

雷赛控制卡以其接口简单、编程便捷、性价比高的特点,成为中小自动化设备的理想选择。使用时需注意:硬件连接需注重抗干扰设计(屏蔽线、接地);软件开发需合理配置运动参数(脉冲当量、加减速);调试阶段需充分利用 DMCStudio 工具验证性能。通过本文介绍的方法,可快速实现从硬件搭建到复杂运动控制的全流程开发,满足雕刻、切割、装配等多场景的自动化需求。

品质生活,极致编程。 3781

基于阻抗控制的工业机器人轨迹跟踪系统 Simulink/Simscape 仿真

  本文是对文献 [1] 的复现。 关于阻抗控制   阻抗控制的提出可以追溯到 1984 年,Neville Hogan 在 American Control Conference 发表的 Impedance Control:An Approach to Manipulation ,次年,Hogan N. 又在 Journal of Dynamic Systems Measurement and Control-transactions of The Asme 发表了阻抗控制经典三部曲 —— Imped.

-贝塔- 的博客 2万+

抽象主义人工智能和生命体AI的内在联系

彭军辉提出“抽象主义人工智能”作为第四大AI范式,批判符号主义、连接主义、行为主义剥离生命本体的误区,确立“智能是现象,生命是本体”的核心公理。主张原生智能必须依附具备存续诉求的生命本体,提出“生命体AI理论”,构建七层演化阶梯与自闭环工程框架,强调智能不可跳过底层存续机制直接模拟语言。该体系为生成式AI提供科学边界,推动从“仿真表象”向“原生智能”的范式跃迁,开辟硅基人工生命新赛道。

抽象主义人工智能和生命体AI 206

Linux性能调优之CPU时间分布(亲和性、带宽、权重)配置

带宽权重亲和性进程级别Cgroup-v2服务级别systyemd我所渴求的,無非是將心中脫穎語出的本性付諸生活,為何竟如此艱難呢 ------赫尔曼·黑塞《德米安》

山河已无恙 1717

OpenAI Control Room 架构拆解:给企业 AI Agent 加一套可审计的急停系统

OpenAI 公开了用于追踪和停止研究模型工作负载的 Control Room。本文拆解 job_id、推理前鉴权、35 秒撤销窗口与 Fail Closed 设计,并给出企业 AI Agent 急停控制面的数据模型、FastAPI 参考实现、缓存撤销策略和故障演练清单。

jerry764的博客 340

龙芯2K1000 OTG双模驱动深度解析

是Synopsys公司开发的USB 2.0 OTG控制器IP核,龙芯2K1000内置该IP,作为唯一的OTG控制器(PCI Dev4 Func0),物理基地址0x40000000。核心能力:同时支持Host主机模式(接U盘/键盘)、Device设备模式(虚拟串口ttyGS0)、OTG双模自动切换速率驱动路径:内核源码PMON:启用otg0设备树内核:开启HOSTPERIPHERALG_SERIAL启动参数脚本:sysfs角色切换+开机自启Host按以上配置,龙芯2K1000 DWC2 OTG可完美实现。

weixin_44932366的博客 377

深度学习工具包 DeepLearningToolbox

深度学习工具包Deprecation notice.-----This toolbox is outdated and no longer maintained.There are much better tools available for deep learning than this toolbox, e.g. [Theano](http://deeplearning.net/software/theano/), [torch](http://torch.ch/) or [tensorflow](http://www.tensorflow.org/)I would suggest you use one of the tools mentioned above rather than use this toolbox.Best, Rasmus.DeepLearnToolbox================A Matlab toolbox for Deep Learning.Deep Learning is a new subfield of machine learning that focuses on learning deep hierarchical models of data.It is inspired by the human brain's apparent deep (layered, hierarchical) architecture.A good overview of the theory of Deep Learning theory is[Learning Deep Architectures for AI](http://www.iro.umontreal.ca/~bengioy/papers/ftml_book.pdf)For a more informal introduction, see the following videos by Geoffrey Hinton and Andrew Ng.* [The Next Generation of Neural Networks](http://www.youtube.com/watch?v=AyzOUbkUf3M) (Hinton, 2007)* [Recent Developments in Deep Learning](http://www.youtube.com/watch?v=VdIURAu1-aU) (Hinton, 2010)* [Unsupervised Feature Learning and Deep Learning](http://www.youtube.com/watch?v=ZmNOAtZIgIk) (Ng, 2011)If you use this toolbox in your research please cite [Prediction as a candidate for learning deep hierarchical models of data](http://www2.imm.dtu.dk/pubdb/views/publication_details.php?id=6284)```@MASTERSTHESIS\{IMM2012-06284, author = "R. B. Palm", title = "Prediction as a candidate for learning deep hierarchical models of data", year = "2012",}```Contact: rasmusbergpalm at gmail dot comDirectories included in the toolbox-----------------------------------`NN/` - A library for Feedforward Backpropagation Neural Networks`CNN/` - A library for Convolutional Neural Networks`DBN/` - A library for Deep Belief Networks`SAE/` - A library for Stacked Auto-Encoders`CAE/` - A library for Convolutional Auto-Encoders`util/` - Utility functions used by the libraries`data/` - Data used by the examples`tests/` - unit tests to verify toolbox is workingFor references on each library check REFS.mdSetup-----1. Download.2. addpath(genpath('DeepLearnToolbox'));Example: Deep Belief Network---------------------```matlabfunction test_example_DBNload mnist_uint8;train_x = double(train_x) / 255;test_x = double(test_x) / 255;train_y = double(train_y);test_y = double(test_y);%% ex1 train a 100 hidden unit RBM and visualize its weightsrand('state',0)dbn.sizes = [100];opts.numepochs = 1;opts.batchsize = 100;opts.momentum = 0;opts.alpha = 1;dbn = dbnsetup(dbn, train_x, opts);dbn = dbntrain(dbn, train_x, opts);figure; visualize(dbn.rbm{1}.W'); % Visualize the RBM weights%% ex2 train a 100-100 hidden unit DBN and use its weights to initialize a NNrand('state',0)%train dbndbn.sizes = [100 100];opts.numepochs = 1;opts.batchsize = 100;opts.momentum = 0;opts.alpha = 1;dbn = dbnsetup(dbn, train_x, opts);dbn = dbntrain(dbn, train_x, opts);%unfold dbn to nnnn = dbnunfoldtonn(dbn, 10);nn.activation_function = 'sigm';%train nnopts.numepochs = 1;opts.batchsize = 100;nn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.10, 'Too big error');```Example: Stacked Auto-Encoders---------------------```matlabfunction test_example_SAEload mnist_uint8;train_x = double(train_x)/255;test_x = double(test_x)/255;train_y = double(train_y);test_y = double(test_y);%% ex1 train a 100 hidden unit SDAE and use it to initialize a FFNN% Setup and train a stacked denoising autoencoder (SDAE)rand('state',0)sae = saesetup([784 100]);sae.ae{1}.activation_function = 'sigm';sae.ae{1}.learningRate = 1;sae.ae{1}.inputZeroMaskedFraction = 0.5;opts.numepochs = 1;opts.batchsize = 100;sae = saetrain(sae, train_x, opts);visualize(sae.ae{1}.W{1}(:,2:end)')% Use the SDAE to initialize a FFNNnn = nnsetup([784 100 10]);nn.activation_function = 'sigm';nn.learningRate = 1;nn.W{1} = sae.ae{1}.W{1};% Train the FFNNopts.numepochs = 1;opts.batchsize = 100;nn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.16, 'Too big error');```Example: Convolutional Neural Nets---------------------```matlabfunction test_example_CNNload mnist_uint8;train_x = double(reshape(train_x',28,28,60000))/255;test_x = double(reshape(test_x',28,28,10000))/255;train_y = double(train_y');test_y = double(test_y');%% ex1 Train a 6c-2s-12c-2s Convolutional neural network %will run 1 epoch in about 200 second and get around 11% error. %With 100 epochs you'll get around 1.2% errorrand('state',0)cnn.layers = { struct('type', 'i') %input layer struct('type', 'c', 'outputmaps', 6, 'kernelsize', 5) %convolution layer struct('type', 's', 'scale', 2) %sub sampling layer struct('type', 'c', 'outputmaps', 12, 'kernelsize', 5) %convolution layer struct('type', 's', 'scale', 2) %subsampling layer};cnn = cnnsetup(cnn, train_x, train_y);opts.alpha = 1;opts.batchsize = 50;opts.numepochs = 1;cnn = cnntrain(cnn, train_x, train_y, opts);[er, bad] = cnntest(cnn, test_x, test_y);%plot mean squared errorfigure; plot(cnn.rL);assert(er<0.12, 'Too big error');```Example: Neural Networks---------------------```matlabfunction test_example_NNload mnist_uint8;train_x = double(train_x) / 255;test_x = double(test_x) / 255;train_y = double(train_y);test_y = double(test_y);% normalize[train_x, mu, sigma] = zscore(train_x);test_x = normalize(test_x, mu, sigma);%% ex1 vanilla neural netrand('state',0)nn = nnsetup([784 100 10]);opts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samples[nn, L] = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.08, 'Too big error');%% ex2 neural net with L2 weight decayrand('state',0)nn = nnsetup([784 100 10]);nn.weightPenaltyL2 = 1e-4; % L2 weight decayopts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samplesnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex3 neural net with dropoutrand('state',0)nn = nnsetup([784 100 10]);nn.dropoutFraction = 0.5; % Dropout fraction opts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samplesnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex4 neural net with sigmoid activation functionrand('state',0)nn = nnsetup([784 100 10]);nn.activation_function = 'sigm'; % Sigmoid activation functionnn.learningRate = 1; % Sigm require a lower learning rateopts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samplesnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex5 plotting functionalityrand('state',0)nn = nnsetup([784 20 10]);opts.numepochs = 5; % Number of full sweeps through datann.output = 'softmax'; % use softmax outputopts.batchsize = 1000; % Take a mean gradient step over this many samplesopts.plot = 1; % enable plottingnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex6 neural net with sigmoid activation and plotting of validation and training error% split training data into training and validation datavx = train_x(1:10000,:);tx = train_x(10001:end,:);vy = train_y(1:10000,:);ty = train_y(10001:end,:);rand('state',0)nn = nnsetup([784 20 10]); nn.output = 'softmax'; % use softmax outputopts.numepochs = 5; % Number of full sweeps through dataopts.batchsize = 1000; % Take a mean gradient step over this many samplesopts.plot = 1; % enable plottingnn = nntrain(nn, tx, ty, opts, vx, vy); % nntrain takes validation set as last two arguments (optionally)[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');```[![Bitdeli Badge](https://d2weczhvl823v0.cloudfront.net/rasmusbergpalm/deeplearntoolbox/trend.png)](https://bitdeli.com/free "Bitdeli Badge")

OpenClaw 深度调查:把 AI 助手做成你自己的控制平面

对 openclaw/openclaw 的完整技术尽调:Gateway 架构、Agent 运行时、工具与沙箱、30 个消息通道、181 个插件、安全信任边界、工程代价与采用建议。17 章 / 10 张图,全部数字标注口径,数据截至 2026-10-31。

厚积薄发 349

lwip-2.0.3

lwip是瑞典计算机科学院(SICS)的Adam Dunkels 开发的一个小型开源的TCP/IP协议栈。实现的重点是在保持TCP协议主要功能的基础上减少对RAM 的占用。

AI赋能智慧港口建设|造形家快速生成港口全要素三维场景

随着智慧港口、港口数字孪生项目的持续推进,港口三维场景的需求日益旺盛。传统港口三维建模,往往需要航拍倾斜摄影采集数据,再人工搭建吊机、岸桥、船舶、集装箱等大量工业设施模型,项目周期长、人力成本高,修改迭代繁琐,制约港口可视化项目落地效率。国产AI三维建模工具,依托卫星影像 AI 识别能力,无需航拍作业,一键快速搭建港口完整三维场景,为港口数字孪生、方案汇报、投标演示提供高效的场景解决方案。

m0_69539167的博客 84

efm32 入门.pdf

本PDF共计22页,介绍的是操作驱动API,下载与否自己斟酌。

从设备增点到治理闭环:老旧小区物业数智化落地实践

系统连接合同审批、银行流水、收费记录与项目台账,执行到期提醒、收入核验、异常标色和月度对账,让每一笔资金来源明确、去向可查。针对设施老旧、诉求响应慢、公共收益不透明等问题,津南以6类小区为试点场景,统一建设物业数字化底座,关联楼栋、单元、房屋、业主、空间、设施设备、工单、公共收益等核心数据。系统采用统一编码管理,将摄像头、路灯、井盖、道路、绿化等设施纳入资产台账,解决纸质记录更新滞后、故障责任难追溯的问题。基于历史工单建立分类模型和知识库,辅助识别照明、给排水、电梯、环境秩序、秩序维护等问题;

wuyeshuzhihua的博客 157

企业微信iPad协议详解:接口接入、实时消息同步与语音转码实践

深入解析企业微信iPad协议的技术架构与开发思路,涵盖API接口接入、长连接通信、实时消息同步、Silk语音转码、异常重连及高并发优化,帮助开发者了解企业微信自动化系统的设计与实现。

欢迎来到祖安! 511

Product Hunt 热榜 2026-11-13:Plurai Simulation、YouArt Desktop App、Omnigraph

2026-11-13 Product Hunt 新产品精选:Plurai Simulation、YouArt Desktop App、Omnigraph。汇总 15 款产品的中文介绍、功能关键词和产品链接。

LaughingZhu的博客 614

【AI Agent面试系列】AI Agent可观测、评测、安全相关的知识点

之前的文章介绍了和,本文主要介绍AI Agent的测评、可观测等知识。

405

Docling 原理解读:是什么、为什么、怎么样

从论文视角拆解 Docling 的整体架构、DoclingDocument 数据模型、PDF 预处理、版面分析、OCR、TableFormer 双解码器、阅读顺序与装配流程,并解释版面框与 OCR 框的区别。

PLANTTHESON的博客 401

WorkBuddy 实战:一句话需求 → 七屏可交互原型,户外徒步 App 的 AI 设计全流程

WorkBuddy 实战:一句话需求 → 七屏可交互原型,户外徒步 App 的 AI 设计全流程,跟着博主一起来还原整个过程。

夜郎King的博客 5156

从认识人工智能到理解机器学习:我的 AI 入门第一课

正式开始学习人工智能之前,我对 AI 的认识其实比较模糊:知道 ChatGPT、知道大模型,也知道现在很多岗位都在谈 AI,但如果有人问我“人工智能到底是什么”“机器是怎么学会识别一只猫的”,我未必能从头到尾讲清楚。上午的课程更偏向行业全景:人工智能的发展、研究方向、NLP、岗位分工、学习路线和 AI Coding 工具。下午的课程更偏向技术基础:机器学习的原理、有监督学习、深度学习、模型训练和评估方法。

chh563的博客 564

企业宣传片发前如何批量排查并遮挡视频里的敏感信息

做企业视频的人都会碰到一个具体问题:实拍画面里混进了不该出现的信息。办公室背景里有贴着客户信息的白板,采访对象的授权书边缘露出一串身份证号码,街采素材里路人面孔和车牌号直接入镜。人眼在时间线上扫视,会漏掉快速闪过的帧、画面边缘的小字、以及暗光区域里的证件号码。用抹影AI完成车牌、人脸和证件区域批量检测后,又手动补充了两个时间段的屏幕遮罩,最终导出时画面已无合规隐患。重点关注快速运动、遮挡、低光照条件下的目标,这些是漏检的高发区域。导入视频后选择要检测的目标类型(人脸、车牌、证件号、二维码等)。

Mysssone的博客 181

面向工业对讲的双通道声学 DSP 模组 A-59(NR2050-A)方案解析与工程落地案例

A-59模组以NR2050-A DSP为核心,37.5×16mm邮票孔封装,支持-40℃~85℃工业温域,兼容A-06/A-09/A-29系列,可低成本替换存量设备。通过固件切换实现三种模式:模式一(低改版成本,适配矿井、电梯);模式二(最优AEC,用于医院呼叫);模式三(双波束独立输出,支持双人会议语音分离)。集成AEC、ENC、BF与AGC算法,有效抑制回声、噪声与串扰,解决对讲系统核心声学难题。选型需注意参考信号路径、麦间距与温度匹配,建议先做声学测试再量产。

R190_6589的博客 371
上一篇: ISOM 5720 ERP & Enterprise Systems
下一篇: ELE000042C Programming and Digital Interfacing
wx____codehelp
博客等级 码龄2年 538粉丝 99原创
评论
成就一亿技术人!
拼手气红包6.0元
还能输入1000个字符
 
 条评论被折叠 查看
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付:10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值