1 00:00:06,137 --> 00:00:07,970 SARAH HANSEN: Could you tell us a little bit 2 00:00:07,970 --> 00:00:12,320 about the history and development of 18.065, 3 00:00:12,320 --> 00:00:13,070 this new course? 4 00:00:13,070 --> 00:00:14,400 GILBERT STRANG: Yes, OK. 5 00:00:14,400 --> 00:00:21,540 So this was my adventure into the subject of deep learning. 6 00:00:21,540 --> 00:00:24,785 So that's a special part of machine learning, 7 00:00:24,785 --> 00:00:26,380 a highly important part. 8 00:00:26,380 --> 00:00:29,780 It's in the newspaper all the time. 9 00:00:29,780 --> 00:00:37,800 And students are graduating knowing about the deep learning 10 00:00:37,800 --> 00:00:39,810 and getting good jobs. 11 00:00:39,810 --> 00:00:43,110 And it's really amazing. 12 00:00:43,110 --> 00:00:46,110 So and then the other beautiful part 13 00:00:46,110 --> 00:00:50,420 is that it depends so heavily on linear algebra. 14 00:00:50,420 --> 00:00:52,920 SARAH HANSEN: You've been known to say, "I certainly learned 15 00:00:52,920 --> 00:00:55,920 that projects are far better than exams. 16 00:00:55,920 --> 00:00:59,580 Students ask their own questions and write their own programs. 17 00:00:59,580 --> 00:01:01,972 From now on, projects." 18 00:01:01,972 --> 00:01:03,180 GILBERT STRANG: That's right. 19 00:01:03,180 --> 00:01:03,880 Yeah. 20 00:01:03,880 --> 00:01:09,090 Sort of late to learn this because with the 18.06 21 00:01:09,090 --> 00:01:12,300 linear algebra, totally conventional course 22 00:01:12,300 --> 00:01:18,280 with maybe three exams during the semester and a final exam. 23 00:01:18,280 --> 00:01:20,370 And maybe that's appropriate. 24 00:01:20,370 --> 00:01:22,260 But it's changing. 25 00:01:22,260 --> 00:01:26,070 And then with this new course, where the computer's involved, 26 00:01:26,070 --> 00:01:26,880 and you-- 27 00:01:26,880 --> 00:01:30,200 for example, so I ask everybody to do a project. 28 00:01:30,200 --> 00:01:31,960 There is no final exam. 29 00:01:31,960 --> 00:01:33,580 Actually, there's no exam at all. 30 00:01:33,580 --> 00:01:35,310 I shouldn't like to say this. 31 00:01:35,310 --> 00:01:41,670 But it's really what the subject is is having an idea of how-- 32 00:01:41,670 --> 00:01:45,987 OK, I'll use deep learning for something. 33 00:01:45,987 --> 00:01:47,820 SARAH HANSEN: What did that feel like to try 34 00:01:47,820 --> 00:01:50,302 something new pedagogically? 35 00:01:50,302 --> 00:01:51,510 GILBERT STRANG: Oh, it's fun. 36 00:01:51,510 --> 00:01:53,670 You know, I like teaching. 37 00:01:53,670 --> 00:01:56,220 And this is a subject where students 38 00:01:56,220 --> 00:01:58,680 just come from everywhere. 39 00:01:58,680 --> 00:02:01,050 Because they know what stuff to learn. 40 00:02:01,050 --> 00:02:02,550 And they've heard about it. 41 00:02:02,550 --> 00:02:05,040 And they-- some of them know more than me. 42 00:02:05,040 --> 00:02:09,680 And then those students write even better projects. 43 00:02:09,680 --> 00:02:18,910 Yeah, it's just-- so I do the lectures for the first three 44 00:02:18,910 --> 00:02:20,340 quarters of the course. 45 00:02:20,340 --> 00:02:23,850 And then I try to get them to present, 46 00:02:23,850 --> 00:02:26,250 which is a great experience for them, 47 00:02:26,250 --> 00:02:30,750 though it takes a little urging to get them. 48 00:02:30,750 --> 00:02:35,760 But, yeah, yeah, it's really just wonderful. 49 00:02:35,760 --> 00:02:37,260 SARAH HANSEN: What insights have you 50 00:02:37,260 --> 00:02:41,380 gained about having more of a student-led course 51 00:02:41,380 --> 00:02:42,990 and a project-based course? 52 00:02:42,990 --> 00:02:44,420 Anything that other-- 53 00:02:44,420 --> 00:02:47,490 GILBERT STRANG: You realize, slowly but finally, 54 00:02:47,490 --> 00:02:50,850 that that's how people learn, by doing. 55 00:02:50,850 --> 00:02:54,150 That you couldn't give them a better way 56 00:02:54,150 --> 00:02:57,530 to learn than to create a project. 57 00:02:57,530 --> 00:03:00,480 Usually, it's on some topic they know about 58 00:03:00,480 --> 00:03:04,710 or they're interested in, like, how do you 59 00:03:04,710 --> 00:03:07,800 find a criminal in a bunch of people? 60 00:03:11,060 --> 00:03:17,740 Yeah, it's just a very effective way to learn. 61 00:03:17,740 --> 00:03:22,350 And it's something that gets remembered, where 62 00:03:22,350 --> 00:03:24,630 doing exam questions that I might make up, 63 00:03:24,630 --> 00:03:28,500 sort of mathy questions, I don't know if that's 64 00:03:28,500 --> 00:03:30,150 remembered 10 years later. 65 00:03:30,150 --> 00:03:32,768 But I think people's projects are. 66 00:03:32,768 --> 00:03:34,560 SARAH HANSEN: Is there anything you learned 67 00:03:34,560 --> 00:03:36,940 from teaching it this way? 68 00:03:36,940 --> 00:03:40,590 Like something that maybe went wrong in the logistics 69 00:03:40,590 --> 00:03:43,560 of facilitating these projects that next time you 70 00:03:43,560 --> 00:03:44,770 want to do differently? 71 00:03:44,770 --> 00:03:45,895 GILBERT STRANG: Well, sure. 72 00:03:48,600 --> 00:03:53,490 I didn't have any idea what to expect really. 73 00:03:53,490 --> 00:03:55,290 And maybe the students didn't either. 74 00:03:55,290 --> 00:03:56,790 They said, what's a project? 75 00:03:56,790 --> 00:04:00,300 Well, one student had said unwisely, 76 00:04:00,300 --> 00:04:03,870 when are the projects due? 77 00:04:03,870 --> 00:04:05,380 I thought, what's that? 78 00:04:05,380 --> 00:04:07,740 I mean, I hadn't even thought about projects. 79 00:04:07,740 --> 00:04:09,193 So I was like, OK. 80 00:04:09,193 --> 00:04:09,900 All right. 81 00:04:09,900 --> 00:04:11,220 You asked for it. 82 00:04:11,220 --> 00:04:15,210 So we decided that the last-- 83 00:04:15,210 --> 00:04:17,329 the end of the semester, the final day 84 00:04:17,329 --> 00:04:21,959 of class, which is two weeks away, 85 00:04:21,959 --> 00:04:24,270 projects are due to come in. 86 00:04:24,270 --> 00:04:27,570 And then some of the class is able, has a chance, 87 00:04:27,570 --> 00:04:32,640 to present their project in the last weeks, but not everybody 88 00:04:32,640 --> 00:04:34,560 because it's a big class. 89 00:04:34,560 --> 00:04:37,710 So it starts with each student or each group-- 90 00:04:37,710 --> 00:04:41,580 it could be two or three students together-- 91 00:04:41,580 --> 00:04:44,880 sends me an email about their plan. 92 00:04:44,880 --> 00:04:46,470 And I respond. 93 00:04:46,470 --> 00:04:48,480 I usually respond, wonderful. 94 00:04:48,480 --> 00:04:52,260 And maybe I have an idea of a reference or two 95 00:04:52,260 --> 00:04:54,630 that they could look at. 96 00:04:54,630 --> 00:04:58,270 But and then they just do it. 97 00:04:58,270 --> 00:05:02,070 Yeah, it's really very nice. 98 00:05:02,070 --> 00:05:04,772 SARAH HANSEN: Do you give them any feedback along the way? 99 00:05:04,772 --> 00:05:06,480 GILBERT STRANG: If they ask for it, yeah. 100 00:05:06,480 --> 00:05:08,100 And if I-- you know, usually they'll 101 00:05:08,100 --> 00:05:11,530 know more about their subject than me. 102 00:05:11,530 --> 00:05:16,500 But maybe what they learn also is presenting. 103 00:05:16,500 --> 00:05:19,570 That's an important thing. 104 00:05:19,570 --> 00:05:24,140 So it's really just more-- 105 00:05:24,140 --> 00:05:29,880 it's richer than taking an exam. 106 00:05:29,880 --> 00:05:33,240 Well, the viewer may think, OK, Professor Strang 107 00:05:33,240 --> 00:05:34,680 says no more exams. 108 00:05:34,680 --> 00:05:35,720 I don't know if-- 109 00:05:35,720 --> 00:05:37,770 so don't quote me on that, please. 110 00:05:37,770 --> 00:05:39,840 SARAH HANSEN: OK. 111 00:05:39,840 --> 00:05:44,400 In 18.065, in one of the videos, you 112 00:05:44,400 --> 00:05:47,750 talk about grading students' work. 113 00:05:47,750 --> 00:05:50,880 And you tell them that, although this 114 00:05:50,880 --> 00:05:53,080 is important to grade their work, 115 00:05:53,080 --> 00:05:55,170 it's not your main concern. 116 00:05:55,170 --> 00:05:58,013 That your main concern is actually learning with them. 117 00:05:58,013 --> 00:05:58,930 GILBERT STRANG: Right. 118 00:05:58,930 --> 00:06:01,340 This is what I want to say the most. 119 00:06:01,340 --> 00:06:04,560 And I say it to every class I teach 120 00:06:04,560 --> 00:06:08,490 near the start of the semester. 121 00:06:08,490 --> 00:06:12,840 My feeling about what my job is is to teach you things 122 00:06:12,840 --> 00:06:17,340 or to join with you in learning things, as has happened today. 123 00:06:17,340 --> 00:06:18,930 It's not to grade you. 124 00:06:18,930 --> 00:06:23,390 So typically, the first few days of class, these guys 125 00:06:23,390 --> 00:06:27,070 ask, you know, what's the class average going to be? 126 00:06:27,070 --> 00:06:28,770 How are we going to be graded? 127 00:06:28,770 --> 00:06:31,260 I don't have any answers for that stuff. 128 00:06:31,260 --> 00:06:35,130 So I say what is totally true, that I'm not-- 129 00:06:35,130 --> 00:06:37,470 I don't feel my main job is to grade them. 130 00:06:40,050 --> 00:06:45,510 That my job is to teach them or learn with them. 131 00:06:45,510 --> 00:06:48,810 And that's what I continue to do. 132 00:06:48,810 --> 00:06:52,080 And gradually, they begin to believe. 133 00:06:52,080 --> 00:06:56,940 You know, at the beginning, they still think, OK, but he's 134 00:06:56,940 --> 00:07:01,050 got to give me a B, or a C, or an A. 135 00:07:01,050 --> 00:07:09,010 But really, that's not what 18.065 is about, a grade. 136 00:07:09,010 --> 00:07:09,690 It's just not. 137 00:07:12,940 --> 00:07:16,090 SARAH HANSEN: What advice do you have for new professors 138 00:07:16,090 --> 00:07:17,590 starting out in teaching? 139 00:07:20,110 --> 00:07:25,210 GILBERT STRANG: Well, probably this interview 140 00:07:25,210 --> 00:07:27,670 has expressed most of the thoughts 141 00:07:27,670 --> 00:07:31,870 I have about in the class. 142 00:07:31,870 --> 00:07:33,670 Yeah, yeah. 143 00:07:33,670 --> 00:07:39,670 Use big chalk, especially if it's a large class. 144 00:07:39,670 --> 00:07:42,010 It helps your writing get-- 145 00:07:42,010 --> 00:07:44,530 your writing looks impressively level, even, 146 00:07:44,530 --> 00:07:46,590 because of the chalk. 147 00:07:46,590 --> 00:07:49,180 And don't rush. 148 00:07:49,180 --> 00:07:52,420 And don't think you have to cover everything. 149 00:07:52,420 --> 00:07:55,550 Just stay with the class. 150 00:07:55,550 --> 00:07:58,200 Yeah. 151 00:07:58,200 --> 00:07:58,950 Yeah. 152 00:07:58,950 --> 00:08:04,590 Well, so again, it's the best job possible.