COMP0197代寫、Python程序設(shè)計(jì)代做

            時(shí)間:2024-03-22  來源:  作者: 我要糾錯(cuò)



            COMP0197 CW1
            1
            COMP0197: Applied Deep Learning
            Assessed Component 1 (Individual Coursework) 2023-24
            Submission before 16:00 (UK time), 21st March 2024 (subject to change), on Moodle
            Introduction
            This is the first of two assessed coursework. This coursework accounts for 50% of the module with three
            independent tasks, and for each task, a task script needs to be submitted with other supporting files and
            data. No separate written report is required.
            There are hyperlinks in the document for further reference. Throughout this document, various parts of
            the text are highlighted, for example:
            The aim of the coursework is to develop and assess your ability a) to understand the technical and
            scientific concepts behind deep learning theory and applications, b) to research the relevant methodology
            and implementation details of the topic, and c) to develop the numerical algorithms in Python and one of
            the deep learning libraries TensorFlow and PyTorch. Although the assessment does not place emphasis
            on coding skills and advanced software development techniques, basic programming knowledge will be
            taken into account, such as the correct use of NumPy arrays, tensors – as opposed to, for example,
            unnecessary for-loops, sufficient commenting and consistent code format. Up to [20%] of the relevant
            marks may be deducted for substandard programming practice.
            Do NOT use this document for any other purposes or share with others. The coursework remains UCL
            property as teaching materials. You may be risking breaching intellectual property regulations and/or
            academic misconduct, if you publish the details of the coursework or distribute this further.
            Conda environment and Python packages
            No external code (open-source or not) should be used for the purpose of this coursework. No other
            packages should be used, unless specified and installed within the conda environment below. This will be
            assessed by running the submitted code on the markers’ computers, within a conda environment created
            as follows, for either TensorFlow or PyTorch. Make sure your OS is up-to-date to minimise potential
            compatibility issues.
            conda create -n comp0197-cw1-tf pillow=10.2 pip=19.3 && conda activate comp0197-cw1-tf && pip
            install tensorflow==2.13
            conda create -n comp0197-cw1-pt -c pytorch python=3.12 pytorch=2.2 torchvision=0.17
            Class names are highlighted for those mandatory classes that should be found in your submitted code.
            Function names are highlighted for those mandatory functions that should be found in your submitted
            code.
            Printed messages on terminal when running the task scripts.
            Visualisation saved into PNG files with task scripts.
            [5]: square brackets indicate marks, with total marks being 100, for 50% of the module assessment.
            “filepath.ext”: quotation marks indicate the names of files or folders.
            commands: commands run on bash or Python terminals, given context.
            COMP0197 CW1
            2
            Use one of the two for your coursework and indicate with your submitted folder name, “cw1-tf” or “cw1-
            pt”. Use the command conda list -n comp0197-cw1-xx to see the available libraries for this coursework
            (“xx” is either “tf” or “pt”). You can choose to use either TensorFlow or PyTorch, but NOT both of them in
            this coursework, as it is designed to have a balanced difficulties from different tasks. [100%] of the
            relevant marks may be deducted for using external code.
            Working directory and task script
            Each task should have a task folder, named as “task1”, “task2” and “task3”. A Python task script should
            be a file named as “task.py”, such that the script can be executed on a bash terminal when the task folder
            is used as the current/working directory, within the conda environment described above:
            python task.py
            It is the individual’s responsibility to make sure the submitted task scripts can run, in the above-specified
            conda environment. If using data/code available in module tutorials, copies or otherwise automated links
            need to be provided to ensure a standalone executability of the submitted code. Care needs to be taken
            in correct use of relative paths, as it was found to be one of the most common issues in the past. Jupyter
            Notebook files are NOT allowed. Up to [100%] of the relevant marks may be deducted if no runnable task
            script is found.
            Printing and visualisation
            Summarising and communicating your implementation and quantitative results is being assessed as part
            of the module learning outcome. Each task specifies relevant information and messages to be printed on
            terminal, which may contain description, quantitative summary and brief remarks. The printed messages
            are expected to be concise, accurate and clear.
            When the task requires visualising results (usually in the form of image), the code should save the results
            into a PNG file in the respective working directory. These PNG files should be submitted with the code,
            although they can be generated by the code as well. Please see examples in the module repository using
            Pillow. Please note that matplotlib cannot be used in the task scripts but may be a good tool during
            development. Up to [50%] of the relevant marks maybe deducted if this is not followed.
            Design your code
            The functions/classes/files/messages highlighted (see Introduction) are expected to be found in your
            submitted code, along with the task scripts. If not specifically required, you have freedom in designing
            your own code, for example, data type, variables, functions, scripts, modules, classes and/or extra results
            for discussion. These will be assessed for complementing your work but not for design aspects.
            The checklist
            This is a list of things that help you to check before submission.
            ✓ The coursework will be submitted as a single “cw1-xx” folder, compressed as a single zip file.
            ✓ Under your “cw1-xx” folder, you should have three subfolders, “task1”, “task2” and “task3”.
            ✓ The task scripts run without needing any additional files, data or customised paths.
            ✓ All the classes and functions colour-coded in this document can be found in the exact names.
            ✓ Check all the functions/classes have a docstring indicating a brief description of its purpose,
            together with data type, size and what-it-is, for each input argument and output.
            COMP0197 CW1
            3
            Task 1 Stochastic Minibatch Gradient Descent for Linear Models
            • Implement a polynomial function polynomial_fun, that takes two input arguments, a weight vector 𝐰
            of size 𝑀 + 1 and an input scalar variable 𝑥, and returns the function value 𝑦. The polynomial_fun
            should be vectorised for multiple pairs of scalar input and output, with the same 𝐰. [5]
            𝑦 = ∑ 𝑤𝑚𝑥
            𝑚
            𝑀
            𝑚=0
            • Using the linear algebra modules in TensorFlow/PyTorch, implement a least square solver for fitting
            the polynomial functions, fit_polynomial_ls, which takes 𝑁 pairs of 𝑥 and target values𝑡 as input, with
            an additional input argument to specify the polynomial degree 𝑀, and returns the optimum weight
            vector 𝐰̂ in least-square sense, i.e. ‖𝑡 − 𝑦‖
            2
            is minimised. [5]
            • Using relevant functions/modules in TensorFlow/PyTorch, implement a stochastic minibatch gradient
            descent algorithm for fitting the polynomial functions, fit_polynomial_sgd, which has the same input
            arguments as fit_polynomial_ls does, with additional two input arguments, learning rate and
            minibatch size. This function also returns the optimum weight vector 𝐰̂. During training, the function
            should report the loss periodically using printed messages. [5]
            • Implement a task script “task.py”, under folder “task1”, performing the following: [15]
            o Use polynomial_fun (𝑀 = 2, 𝐰 = [1,2,3]
            T
            ) to generate a training set and a test set, in the
            form of respectively and uniformly sampled 20 and 10 pairs of 𝑥, 𝑥𝜖[−20, 20], and 𝑡. The
            observed 𝑡 values are obtained by adding Gaussian noise (standard deviation being 0.5) to 𝑦.
            o Use fit_polynomial_ls (𝑀𝜖{2,3,4}) to compute the optimum weight vector 𝐰̂ using the
            training set. For each 𝑀, compute the predicted target values 𝑦̂ for all 𝑥 in both the training
            and test sets.
            o Report, using printed messages, the mean (and standard deviation) in difference a) between
            the observed training data and the underlying “true” polynomial curve; and b) between the
            “LS-predicted” values and the underlying “true” polynomial curve.
            o Use fit_polynomial_sgd (𝑀𝜖{2,3,4}) to optimise the weight vector 𝐰̂ using the training set.
            For each 𝑀, compute the predicted target values 𝑦̂ for all 𝑥 in both the training and test sets.
            o Report, using printed messages, the mean (and standard deviation) in difference between the
            “SGD-predicted” values and the underlying “true” polynomial curve.
            o Compare the accuracy of your implementation using the two methods with ground-truth on
            test set and report the root-mean-square-errors (RMSEs) in both 𝐰 and 𝑦 using printed
            messages.
            o Compare the speed of the two methods and report time spent in fitting/training (in seconds)
            using printed messages.
            • Implement a task script “task1a.py”, under folder “task1”. [10]
            o Experiment how to make 𝑀 a learnable model parameter and using SGD to optimise this more
            flexible model.
            o Report, using printed messages, the optimised 𝑀 value and the mean (and standard deviation) in
            difference between the model-predicted values and the underlying “true” polynomial curve.
            Task 2 A depth-wise separable convolution
            For the purpose of the coursework, the dataset is only split into two, training and test sets.
            COMP0197 CW1
            4
            • Adapt the Image Classification tutorial to use a different network, VisionTransformer. You can choose
            any configuration that is appropriate for this application. [5]
            o TensorFlow version
            o PyTorch version
            • Implement a data augmentation class MixUp, using the mixup algorithm, such that: [10]
            o Inherited from the relevant classes in TensorFlow/PyTorch is recommended but not assessed.
            o The MixUp algorithm can be applied to images and labels in each training iteration.
            o Have an input flag “sampling_method” and appropriate hyperparameters for two options:
            ▪ sampling_method = 1: λ is sampled from a beta distribution as described in the paper.
            ▪ sampling_method = 2: λ is sampled uniformly from a predefined range.
            ▪ The algorithm should be seeded for reproducible results.
            o Visualise your implementation, by saving to a PNG file “mixup.png”, a montage of 16 images
            with randomly augmented images that are about to be fed into network training.
            o Note: the intention of this task is to implement the augmentation class from scratch using
            only TensorFlow/PyTorch basic API functions. Using the built-in data augmentation classes
            may result in losing all relevant marks.
            • Implement a task script “task.py”, under folder “task2”, completing the following: [15]
            o Train a new VisionTransformer classification network with MixUp data augmentation, for
            each of the two sampling methods, with 20 epochs.
            o Save the two trained models and submit your trained models within the task folder.
            o Report the test set performance in terms of classification accuracy versus the epochs.
            o Visualise your results, by saving to a PNG file “result.png”, a montage of 36 test images with
            printed messages clearly indicating the ground-truth and the predicted classes for each.
            Task 3 Ablation Study
            Using the Image Classification tutorial, this task investigates the impact of the following modification to
            the original network. To evaluate a modification, an ablation study can be used by comparing the
            performance before and after the modification.
            • Difference between training with the two λ sampling methods in Task 2.
            • Implement a task script “task.py”, under folder “task3”, completing the following: [30]
            o Random split the data into development set (80%) and holdout test set (20%).
            o Random split the development set into train (90%) and validation sets (10%).
            o Design at least one metric, other than the loss, on validation set, for monitoring during
            training.
            o Train two models using the two different sampling methods.
            o Report a summary of loss values, speed, metric on training and validation.
            o Save and submit these two trained models within the task folder.
            o Report a summary of loss values and the metrics on the holdout test set. Compare the results
            with those obtained during development.
            請(qǐng)加QQ:99515681  郵箱:99515681@qq.com   WX:codehelp

            標(biāo)簽:

            掃一掃在手機(jī)打開當(dāng)前頁(yè)
          1. 上一篇:代做COMP226、代寫solution.R設(shè)計(jì)編程
          2. 下一篇:代做CSMBD21、代寫Java, C/C++, Python編程
          3. 無相關(guān)信息
            昆明生活資訊

            昆明圖文信息
            蝴蝶泉(4A)-大理旅游
            蝴蝶泉(4A)-大理旅游
            油炸竹蟲
            油炸竹蟲
            酸筍煮魚(雞)
            酸筍煮魚(雞)
            竹筒飯
            竹筒飯
            香茅草烤魚
            香茅草烤魚
            檸檬烤魚
            檸檬烤魚
            昆明西山國(guó)家級(jí)風(fēng)景名勝區(qū)
            昆明西山國(guó)家級(jí)風(fēng)景名勝區(qū)
            昆明旅游索道攻略
            昆明旅游索道攻略
          4. 福建中專招生網(wǎng) NBA直播 短信驗(yàn)證碼平臺(tái) 幣安官網(wǎng)下載 WPS下載

            關(guān)于我們 | 打賞支持 | 廣告服務(wù) | 聯(lián)系我們 | 網(wǎng)站地圖 | 免責(zé)聲明 | 幫助中心 | 友情鏈接 |

            Copyright © 2025 kmw.cc Inc. All Rights Reserved. 昆明網(wǎng) 版權(quán)所有
            ICP備06013414號(hào)-3 公安備 42010502001045

            主站蜘蛛池模板: 精品综合一区二区三区| 本免费AV无码专区一区| 午夜视频在线观看一区二区| 国产一区在线mmai| 成人在线观看一区| 一区二区网站在线观看| 国模大尺度视频一区二区| 久久精品岛国av一区二区无码| 久久国产精品无码一区二区三区| 91精品乱码一区二区三区| 精品国产日韩亚洲一区在线| 一区二区三区在线看| 亚洲一区二区三区四区在线观看| 麻豆一区二区在我观看 | 日韩美女在线观看一区| 无码毛片一区二区三区中文字幕 | 美日韩一区二区三区| 一区二区三区www| 国产成人精品一区二区秒拍| 中文字幕AV一区二区三区| 能在线观看的一区二区三区| 国产午夜精品免费一区二区三区| 国产精品合集一区二区三区| 国产一区在线观看免费| 无码人妻久久久一区二区三区| 黑巨人与欧美精品一区| 亚洲色无码一区二区三区| 日产亚洲一区二区三区| 在线播放精品一区二区啪视频| 91久久精一区二区三区大全| 亚洲精品日韩一区二区小说| 中文字幕在线一区二区三区| 亚洲.国产.欧美一区二区三区| 中文字幕一区二区三区有限公司| 国产99视频精品一区| 国产一在线精品一区在线观看| 无码乱码av天堂一区二区| 丝袜人妻一区二区三区网站| 怡红院一区二区在线观看| 老熟妇仑乱视频一区二区| 精品亚洲一区二区三区在线播放|