InceptionWebDesign/problemstatement.html at master · jordanjones243/InceptionWebDesign · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
<!DOCTYPE html>
<html lang="en">

<head>

    <meta charset="utf-8">
    <meta http-equiv="X-UA-Compatible" content="IE=edge">
    <meta name="viewport" content="width=device-width, initial-scale=1">
    <meta name="description" content="">
    <meta name="author" content="">

    <title>Microsoft AzureML - Image Classification Team</title>

    <!-- Bootstrap Core CSS -->
    <link href="css/bootstrap.min.css" rel="stylesheet">

    <!-- Custom CSS -->
    <link href="css/grayscale.css" rel="stylesheet">
</head>
<body>

    <header>
        <h1>Problem Statement - Image Classification</h1>
    </header>

    <p> <strong>Sense of urgency:</strong> As a corporate leader in environmental sustainability, Microsoft is
       seeking new ways of decreasing their carbon footprint through their commitment to environmental
       sustainability, and must seek more opportunities to make their tools more efficient. Along with
       this, they must continue to seek improvements around ML practices such as training datasets so
       their customers can better utilize Azure services. </p>

    <p>Currently, the way in which many tech companies analyze data is unsustainable when compared to
       the rate at which they are collecting the data. Data collection has rapidly increased over the
       years however, the processes that are used to analyze this data on are not advancing at a high
       enough rate to catch up. We find this to be a major drawback for Microsoft's sustainability
       goals as data collection and analysis require high computational hours. To reduce the overall
       ost of operations, the current Microsoft AzureML model needs to be optimized to accurately map
       key efficiencies, power metrics, cost, carbon and accuracy of the model. Our motivation is to
       research and develop a methodology that best fits the Image Classification vertical,
       specifically using the InceptionV3 model as a baseline to create GitHub samples that showcase
       our optimized approach. </p>

    <p>
    Goals of case study: From a general scope, Microsoft wants to have resources that show their new
    AzureML technologies are more cost-effective and use less resources so customers can make more
    informed decisions. Our group specifically is focusing on how we can provide this ability through
    image classification capabilities of AzureML for our stakeholders:
        <ul>
            <li>Create a starting point for customers</li>
            <li>Microsoft wants to integrate cost/power/efficiency metrics into their AzureML software
                to help customers choose the correct tools for what they need.</li>
            <li>Faster training = quicker experimental models can lead to faster turnaround times as
                well as being cheaper. </li>
        </ul>
    </p>

    <!-- Footer -->
    <footer>

    </footer>

</body>

</html>